Production | Tiger Sniff Technology Group

Author | Zhang Xue

Cover cover | IC photo

On the evening of March 26, the Shanghai Stock Exchange issued an announcement stating that it had accepted the application for listing of the Science and Technology Cambrian Technology Co., Ltd. (“Cambrian”) and listed the prospectus declaration.

This is also the first step in the Cambrian science and technology board after the news of receiving guidance on the listing of science and technology board in early December last year.

Public information shows that Cambrian technology was co-founded by Chen Tianshi and his brother Chen Yundi in 2016. Both brothers Chen Tianshi and Chen Yundi have graduated from the Chinese University of Science and Technology Junior Class. Chen Tianshi’s research direction is artificial intelligence algorithms, while Chen Yundi focuses on computer chips.

The Cambrian predecessor was a 10-member academic team of “Exploring the Cross-cutting Field of Processor Architecture and Artificial Intelligence” formed by the Institute of Computing Technology of the Chinese Academy of Sciences in 2008. The name “Cambrian Period” is to hope to use the era of the outbreak of life in geology to compare the future of artificial intelligence.

Although it is not unusual for a science and technology company with a star halo to test the science and technology board, but at this point in time when the stock market is not so good, the Cambrian sprinting the science and technology board still caused industry discussions and attention.

“If the Cambrian Science and Technology Innovation Board is successfully listed, it is very likely that several AI companies will follow suit, and the impact in the industry is still relatively large.” Su Shen, an AI chip industry practitioner, told this Tiger sniffing.

It is reported that the Cambrian, which was established in 2016, has gone through 6 rounds of financing, with a pre-IPO valuation of up to 22.2 billion yuan, which is a veritable AI chip unicorn in China.

The prospectus shows that, in terms of listing standards, the Cambrian chose the second paragraph-it is expected that the total market value after listing will be no less than 1.5 billion yuan, it is planned to issue no more than 40.1 million shares, and it is planned to use raised funds of about 28 100 million yuan.

Continuous losses, is it normal?

Opening the Cambrian prospectus, the first thing to impress is the three sets of loss data:

In 2017, 2018 and 2019, the company’s net profit attributable to ordinary shareholders of the parent company was -380,700,400 yuan, -4,104,65 thousand yuan, and -117,898.56 million yuan.

Obviously, during the reporting periodThe Cambrian did not make a profit. At this point, some even joked that they found the feeling of looking at the Shangtang prospectus that year.

The Cambrian also explained the reasons for the losses. First, the company’s research and development expenditure is large, and the product is still in the market expansion stage; second, the employee incentive share payment during the reporting period is relatively large. Of the two, a larger part of the losses is the former.

Data show that in 2017, 2018, and 2019, the Cambrian confirmed R & D expenses of 29.619 million yuan, 241.118 million yuan, and 540,304,450 yuan respectively. In contrast, the revenues in these three years were 7,843,300 yuan RMB 11,702,500 and RMB 443,385,800.

It is not difficult to see that since the establishment of the Cambrian in 2016, its annual R & D expenses accounted for more than 100% of its revenue.

The huge amount of effort has also brought the corresponding product output.

Since its establishment, the Cambrian has successively launched Cambrian 1A, Cambrian 1H, Cambrian 1M series chips for terminal scenarios, and cloud-based smart acceleration card products based on Siyuan 100 and Siyuan 270 chips. And the edge intelligent acceleration card based on siyuan 220 chip.

However, like every company in the AI ​​field, the output of the product does not bring high returns. At this stage where the AI ​​commercial landing scene is not clear, the Cambrian can’t get rid of the “lose of money”.

Su Shen told Tiger Sniff, “On the basis of existing technology, ordinary chips may invest millions of dollars to get a good product and mass production. But AI chips are not the same. Even if you invest 100 million yuan, you don’t necessarily hear much noise. In addition to the difficulty of production, the repeated debugging of the process, and large-scale production, it really requires a lot of investment. “

“The Cambrian is really short of money, and anxious.” An industry insider who has read the Cambrian prospectus said so.

The person further explained that “different from other small AI companies, the Cambrian has a certain accumulation of technology, but it also needs to be acknowledged that there are still some Cambrian and top AI chip companies Gap, and this gap needs a lot of R & D investment to make up. If investors are not willing to continue to invest, the Cambrian situation will become very embarrassing.

And this guess is alsoThis was confirmed in the prospectus.

It can be seen from the prospectus that the Cambrian will invest the funds raised from the public offering into the development of a new generation of cloud training chips, inference chips and edge-end smart chips and systems.

It should be pointed out that the main force of the Cambrian era is on cloud chips, but it does not have cloud database resources. Sources familiar with the matter said, “Database resources are in the hands of major cloud computing manufacturers, Purchasing a database from these cloud vendors ranges from hundreds of thousands to millions, and still only restricts mobility. If you want more, you have to invest more money, which is a bottomless pit.

There is no doubt that the Cambrian chip design and understanding ability is recognized as the top in the industry, but this cannot avoid the shortcomings of other links.

Another semiconductor industry observer, senior semiconductor investor Kong Jie, also talked about “As a company choosing the AI ​​chip track, the Cambrian needs to endure a low income, low profit, and even a long time. The status of annual losses. “

The Cambrian has a clear understanding of this, and the prospectus reads: “In the future, the Cambrian will continue to lose money and cannot make a profit.”

After all, the fact is that the Cambrian’s working capital now relies more on external financing.

Is it really normal for a sustained loss like the Cambrian?

“It’s normal. Every great company will experience such a difficult time. It doesn’t matter. There are so many investment institutions behind it. It’s good to survive this period of silence.” Kong Jie said so.

But in such a huge loss, suddenly choosing to go public is unreadable to everyone.

Whoever spends money, how about making money?

It is true that before the IPO, the Cambrian was already the darling of capital.

The prospectus details the multiple rounds of funding that Cambrian has carried out since its inception. Its strategic investors include Alibaba, HKUST Xunfei, Chinese Academy of Sciences Venture Capital, Guoxin and other heavyweight companies and capitalists.

For start-up companies, getting capital favors is only the first step, and being able to make money from “money” is the real skill.

IntroductionShao, Cambrian is currently engaged in three major business lines: cloud-based smart chips and accelerator cards, intelligent computing cluster systems, and terminal intelligent processor IP. Among them, in 2017, 2018 and 2019, the company’s main business revenue accounted for 99.38%, 100.00% and 99.99%, respectively.

It can be seen that, unlike the development path of start-ups in the vertical track, the Cambrian is a general route in the business direction. This is the case. Ji is still unclear even now.

In addition, there are two main ways for the Cambrian artificial intelligence chip business model. One is to authorize the processor IP to the chip manufacturer, for example, in cooperation with Huawei; the other is to design and sell the chip itself. For example, Cambrian cloud smart chip launched for Internet companies and other users.

It is worth noting that in 2017 and 2018, Cambrian’s main business revenue was mainly derived from terminal intelligent processor IP license revenue; in 2019, Cambrian only expanded the cloud smart chip and accelerator card business and intelligent computing cluster system business.

So the Cambrian, which seems to walk on two legs, has always been a “cripple” in a way.

When it comes to Cambrian revenue, we have to mention Huawei.

In 2016, the Cambrian, which was just established, launched the first Cambrian IPA 1A on the market.

Subsequently, the Cambrian 1A IP was selected, integrated into the Huawei Kirin 970 processor, and carried on Huawei’s flagship mobile phone Mate 10.

Standing on the shoulders of the giant Huawei, the young breakout Cambrian successfully fired the first shot.

Then, in 2017, the Cambrian launched an upgraded version of the IP Cambrian 1H, which was also integrated into the Kirin 980 processor.

The cooperation with Huawei not only made the Cambrian an instant fame in the industry, but also more practical, it also brought the first bucket of gold to the Cambrian. Data show that the shipments of mobile phones equipped with Kirin 970 and Kirin 980 processors have exceeded 100 million units. In the past two years, thanks to the IP authorization, the Cambrian recorded RMB 7.833 million and RMB 117.052 million, accounting for 98.95% and 99.69% of the main business revenue, respectively.

However, things have changed in 2019, Huawei’s new Kirin 990 chooses to use self-developed NPU instead of Cambrian IP.

The impact of this change on the Cambrian is undoubtedly huge, and it is more intuitively reflected in the revenue. In 2019, the revenue contributed by the company A (Huawei) to the Cambrian was significantly reduced to 6365. Ten thousand yuan, only accounting for 14.34% of the company’s revenue for the year.

Furthermore, in the prospectus, it can be seen more clearly that the Cambrian signed 4 technology license contracts with company A (Huawei), with the contract unit price ranging from 19 million to 28 million. Currently there are still A contract is in the state of being fulfilled.

Anywhere.

Huawei lost the Cambrian Period of Huawei, temporarily abandoned the IP authorization path that it has always relied on, and turned to the expansion of cloud smart chip and accelerator card business and intelligent computing cluster system business in a timely manner.

It may be based on the further maturity of AI cloud chips, or the red background of the Cambrian. In the short term, the Cambrian signed the Intelligent Computing Platform (Phase 2) project of Zhuhai Hengqin New District Management Committee Commerce Bureau. The total contract price was as high as 440 million, which directly brought 200 million revenues to the Cambrian, and also brought new directions for the Cambrian’s lost journey in 2019.

During the same period, the Cambrian also reached related cooperation with Xi’an Chandong Yixiang Technology Service Co., Ltd. and Shanghai Brain Science and Brain-like Research Center.

The Cambrian’s new course has also been recognized by investors in the industry. The person pointed out directly: “I am optimistic about the Cambrian future cloud server market. They can cut a piece of cake in this market because The Cambrian technology level, including the background of the team, and the current leadership of investors, can help the Cambrian have sufficient resources to handle the market. But the most important thing is that this market has no money, and the market itself is not much. , This market will not erupt at least in the next three years.

In other words, even though the Cambrian can gain a place in the cloud market now, the return that this market can bring to the Cambrian is very limited, and the next three years will be three years of continued internal practice.

Looking at this year alone, due to the impact of the epidemic, it is fortunate that the electronics manufacturing industry can return to normal in the first half of the year, and there is nothing that cannot be completed in eight or nine months from the promotion to the application of the products in the AI ​​field. Therefore, in the period when major companies are seeking stability, it is almost impossible for a large number of AI products to be purchased.

Another fact is that the war between big factories has begun to burn, whether it is Huawei or Ali in ChinaStill foreign Intel and Nvidia are staring at this piece of cake desperately.

The Cambrian himself also acknowledges that compared with international large integrated circuit companies such as Nvidia, Intel, and AMD, the company still has a large gap in terms of overall scale, financial strength, R & D reserves, and sales channels. Among domestic enterprises, such as Huawei Hisilicon and other chip design companies are also gradually entering the market, and the company is facing a situation of further intensified market competition.

So after the giants contest, how much can be left to the Cambrian, it is indeed unpredictable.

Back of Cambrian IPO

In the eyes of most people, the Cambrian IPO in the science and technology board has taken a step forward, and it seems to be a good signal for the industry and the Cambrian itself.

Zhu Jingjing, an analyst with the Beijing Municipal Semiconductor Association, told Tiger Sniff: “The Cambrian has been planning to go public. From the perspective of the association, it is very supportive of Cambrian listing. In addition, the Cambrian represents Beijing integrated circuit companies. Sprinting the science and technology board, if the Cambrian can be successfully listed, it will have a good encouragement and driving effect for integrated circuit companies in Beijing, and then increase the enthusiasm of related industries in Beijing.

The so-called layman looks busy, and the insider looks at the doorway.

Many people in the industry believe that there is more crisis on the back of the Cambrian IPO, and the industry still holds a pessimistic attitude about whether the Cambrian can pass the final meeting.

Su Shen said: “In general, there are several options for start-up companies to go public-science and technology board, Hong Kong, Nasdaq, etc. In contrast, science and technology board has the highest requirements for enterprises, the chip industry and This is especially true in the AI ​​industry. The specific requirements in terms of sales are very many, which is also the direct reason that many domestic AI companies and chip companies are discouraged from science and technology boards.

Specifically, the Cambrian period, although the revenue has soared more than 50 times in the past three years, it has actually lost 1.6 billion in three years.

An industry source who did not want to be named speculated, “At this point in time, the science and technology board may relatively relax the listing requirements. After all, some domestic policies continue to drive up the chip industry, or even remain stationary. , Which is a relatively high point in the past. So from the Cambrian perspective, it is also a decision to let go and think carefully.

Unlike what Su Shen was worried about, Kong JieIt is believed that “ Cambrian as a domestic unicorn company of AI chips, listing should not be the purpose, or at least not the current purpose. The reason is that the Cambrian can affect the scope of the entire AI industry. It ’s still very small. Whether it is the number of deployable products, the influence of the product, and the basic profitability, the Cambrian is far from meeting the requirements. “

In his opinion, a mature AI cloud technology company must be a long-term company with a long history. In this mode, he has great potential energy and moves slowly, but once he moves, he will rush to the market with a strong momentum. This is the normal development path of this type of company, not This is now the Cambrian.

At the same time, some people also expressed faint concerns with Tiger Sniff, “If the Cambrian really succeeds in listing, it will promote the trend of short-term cash-out to a certain extent. This is not the case for the Chinese industry and the entire field of artificial intelligence. A good sign. A company that has not reflected the value of the industry at all, and can go public only by telling stories, is not good for this industry. “

On the other hand, Tiger Sniff found that the industry ’s attention to Cambrian listing is so high. In addition to its topical nature, More, it carries the high expectations of the domestic chip industry.

“We really hope that he can become a banner in the field of Chinese chips. The Cambrian potential is recognized by the industry. He has enough algorithms and a good endorsement behind him. At the same time, he There are many related industry backgrounds among the investors, and we really hope to see him grow into a giant in China and even in the world. “A chip veteran said earnestly.

He continued to explain: “Conversely, if it is listed quickly, the Cambrian movement will inevitably be deformed. It is not ruled out that in order to pursue short-term financial benefits, the R & D direction is ignored and more short-term Market plan, the management team will be more inclined to financial management rather than technical management and customer management, especially long-term customer management. “

According to the prospectus declaration, the Cambrian controlling shareholder and actual controller is Chen Tianshi. Before the issuance, it directly held 33.19% of the company’s shares. The company’s wholly-owned Sinotech source was the second largest shareholder of the company, with a shareholding ratio of 18.24%.

In addition, SDIC Fund, Nanjing CMB, Ali Ventures, and HKUST Xunfei are among the shareholders. Before the Issuance, there were 32 individual and institutional shareholders holding direct shares in the Cambrian.

According to the seals that Chen Tianshi and his team left to the industryElephants are mostly down-to-earth and hard-working. Among the Cambrian shareholders, there are more institutions with a state-owned background.

So it seems that the sense of urgency to rush out should not have appeared in the Cambrian. Of course, we have no way of knowing or verifying the reasons behind the Cambrian decision to go public.

It is certain that if Cambrian is successfully landed in the capital market, the Cambrian will become the “first chip of AI chip” in the science and technology board.

That’s because the science and technology board currently needs an AI chip of high quality. Must it be Cambrian?

In the text of the interviewee, Su Shen and Kong Jie are pseudonyms.