Great potential!Another key industry related to the competition between China and the United States has reached a critical moment ...

Surong the winter and usher in overtaking.

Text 丨 Chinese Shang Tao Lue Liu Baizheng

In March 2000, "Baron Weekly" conducted a survey of 207 Internet companies, and then gave a conclusion that 71%of the company was losing money. 51 companies were used up within 12 months, including Amazon Essence

In the article, "Baron Weekly" expressed his concern in one sentence: careful investment.

At the time of the Internet, it was just like AI at this moment.

【Big Tragedy】

On June 30, 2022, the stock price of AI's first shareholder Tang was beheaded and fell below the issue price of Hong Kong dollars. Within a day, about 90 billion Hong Kong dollars (approximately 76.9 billion yuan) were fired. After that, it fell again and again until the edge of HK $ 2 before ushered in a decent rebound.

What is the concept of falling 46.77%within a day?

If the market value is regarded as operating results, it is equivalent to almost half of the hardships of years of struggle within a day.

But the direct reason why Shang Tang is so embarrassing is not what the company's operation has changed. In fact, since listing, the company's revenue has been growing, and the AI ​​industry in the company is still the darling of the times.

Based on various news, the only reason for its abnormal stock price is the restriction of the shares held by investors before listing.

On June 30, due to restrictions on the ban, Shangtang Technology's circulation shares increased from 2%a day ago to 70%, and the daily turnover suddenly increased from the previous approximately HK $ 100 million to HK $ 4.6 billion.

The main role of crazy selling is self -evident.

Although the market is turbulent at the moment, the previous investors have huge profits, and it is understandable for running the road, but this still makes people worry about the prospects of the entire AI industry.

Moreover, this was a bit worried.

Just a few months ago, Shangtang Technology was still the "little sweetness" of investors, but now they seem to don't want to wait a day.

Shangtang Technology is known as the "AI Four Little Dragons". It has a heavenly group with more than 3,000 people and is recognized as "China's strongest team, the top five in the world" in the AI ​​industry. Behind it, there are many investment leaders such as Alibaba, SoftBank, China Structure Adjustment Fund, Chunhua Capital, and Dinghui Investment, and set a single financing record for AI companies.

The achievements of Shangtang Technology have even been recognized by the US government. On the eve of its listing, the US Treasury Department included it on the "Non -SDN Chinese Military Compound Enterprise" and banned American investors from participating in it.

"Use me a hundred a fews of heat to show out thousands of lights." Xu Li, co -founder of Shangtang Technology, tone and bold on the day of the listing.

Recently, during the inspection of the Supreme Leader at the Hong Kong Science Park, he also visited the Shangtang Technology booth, which has set up a branch here and praised its development. In 2018, the leader hosted a symposium for private enterprises. The founder of Shangtang Technology Tang Xiaoou also made a complication of representatives of high -tech enterprises invited to attend.

However, after listing, the stock price rose more than double the Shang Tang, and it was not surprising that the stock price plummeted.

Prior to this, the bleak haze of AI stocks was shrouded in the heads of major AI companies. Today's AI company's stock price performance is also like a large -scale miserable conference.

The earliest AI company Cambrian was the peak shortly after the listing.

Its stock price reached the highest point within three days, and then fell all the way. As of August 2 this year, it has fallen 80%from the highest point.

Cambrian was also favored by investors. After the cooperation with Huawei Hisilicon became famous overnight, it raised five times in four years, and also set a record of domestic listing review. It is known as the "first share of AI chip".

Galing deep pupils listed this year are even more tragic. It broke on the first day of listing on March 17, and once fell nearly 50%deep, and it still fell nearly 24.4%.

However, in a dinner eight years ago, Xu Xiaoping, the founder of Zhenge Fund, believes that Galian deep pupils are valued at least 500 billion US dollars in the future. Although Sequoia Capital partner Shen Nanpeng thought that Xu Xiaoping had drank more nonsense, he also thought that $ 100 billion was more practical.

But now the market value of Guling deep pupils is only 5.672 billion yuan. It seems that in the dinner of that year, Shen Nanpeng actually drank a lot.

Yuncong Technology, who is also "AI Four Little Dragons", has risen by about 40%after listing on May 27 this year, and then there is a bigger wave, but behind it is forced to reduce fundraising funds before listing. Broken arms and issuance of the issue price.

According to the prospectus, Yun raised 3.75 billion yuan from the original science and technology plan, but in the end, it only raised about 1.728 billion yuan.

It can be said that its subsequent stock price rose because it had "declined" before listing. Now, it has doubled to the issue price, and it has begun to approach the issue price.

The shrinkage of the market value of major companies has also made many people think that the bubble of the AI ​​industry has broken, and winter is approaching.

【Embarrassed performance】

Once, when investors watched the AI ​​project, the question was often "how strong is your company's technology? What are the technical guys?"

Now, the problem has become more practical. Cases, cooperative customers, and revenue performance are the core of investors.

In other words, if you want to gain the favor of investors, you can no longer rely on those halo, titles and imagination, but rely on performance and real growth.

The conversion of the problem poked the pain points of AI.

Although the story has been told for many years, almost all AI companies are still in a state of losing money so far, and even caught in the bottleneck of business growth. Shangtang Technology is the company with the strongest strength, the most foundation for business, and the most eye -catching performance. From 2018 to 2021, its revenue has increased from 1.9 billion yuan to 4.7 billion yuan. However, with the high growth of revenue, it is its continuous expansion of losses: from the net loss in 2018 to a net loss of 3.43 billion yuan to 17.17 billion yuan in 2021.

The more revenue, the more you lose. Such embarrassment is not only in Shang Tang.

Today's AI companies are rolling on the road of losing money.

Yitu Technology's cumulative loss of over 7 billion yuan from 2017 to the first half of 2020;

Yuncong Technology's cumulative loss of over 2.6 billion yuan from 2018-2020;

Cumulative losses from Cambrian 2018-2020; 1.656 billion yuan;

The cumulative losses in the first half of the 2018-2021 of Guling Pupil were as high as RMB 618 million.

Although the loss of these companies is much lower than that of Shangtang Technology, behind it, it is even more difficult to look at the revenue of Shang Tang's technology.

In addition to the performance figures, the business progress of each AI company is not optimistic.

First of all, in the current main position of AI commercialization, and the relatively mature computer vision field, head manufacturers have cultivated in various market segments for many years, and the market pattern has become stable.

According to the report of IDC 20121, Shangtang Technology, Yitu Technology, Yuncong Technology, and Guling deep pupils have their own strengths, and each has fierce competition, which can be described as heroes.

Among them, Shangtang Technology's market share accounted for the number one, but it was only 22.2%, and the growth was slow, only 3.8%in a year.

Secondly, in the expansion of emerging fields, specific difficulties have their own difficulties.

According to Shang Tang's financial report, in its major businesses, the fastest growth is smart cities, and more applications of smart cities are AI cameras.

For example, the urban Ark platform deployed in Shenzhen can be used to detect incidents of motorcycle violations of traffic rules, or whether the driver is a seat belt.

However, Haikang, who started video surveillance, has established an absolute advantage in this business field. According to data in 2020, in the global video surveillance market, Haikang's market share has reached 40%.

This also means that in this field, every arch of Shangtang seems to be going to move the mountain.

Yun Cong Technology also mentioned in the prospectus that the company's business development is not smooth sailing, and some investments have been forced to drift.

For example, in 2020, because of the unpredictable chip design, at the same time, the related industries have encountered restrictions, and they have to terminate the "industrialization application of artificial intelligence SoC chip development and combining high accuracy face recognition technology" project, and let the relevant business stagnant stagnation Essence

For example, the popular AI medical field after the epidemic has been considered a new position of AI, but it is far from making money.

Global is also cool and hot, and the progress of the AI ​​industry is not good, not only in China.

The accumulation of technology is as strong as the United States, and it is also facing the problem of losses and burning money for business.

For example, a giant Google in the AI ​​field has established Google health for AI medical care, and also hires a senior person in the medical field David Feinberg as the supervisor. However, according to its first quarter financial report, Google's innovative business, including artificial intelligence DEEPMIND and intelligent medical Verily, is in a state of loss.

In such a situation, Google's health had to lay off 20%in 2021, and it was completely split in 2022.

It is as strong as Google. It has not yet picked the fruit of medical AI, not to mention the domestic AI company that is still in the entrepreneurial period.

【Application Dilemma】

Technology is constantly breaking through, and many things are only theoretically feasible, but the existing application scenarios are fiercely competitive, and the new application scenarios are difficult to break through. This is the main reason for the performance of AI companies.

According to the "2020 China Artificial Intelligence Development Report", from 2011 to 2020, the number of Chinese artificial intelligence patent applications was 390,000, accounting for 74.7%of the global total, which was 8.2 times that of the second United States.

This is the pride of Chinese AI and the capital of future development.

It is a new industry that realizes the hope of overtaking the United States.

But at the moment, many of these patents have almost no use of martial arts.

At present, the Chinese artificial intelligence market is still mainly limited to urban governance and operation, and its proportion of applications has reached 49%, and it is mainly based on face recognition in the security field.

But almost all manufacturers gathered here. There are more monks and less porridge, and their performance is naturally unsatisfactory.

In addition, some scenarios that can be applied today are not so high for technical requirements. For example, the face recognition gate that has been popular for a while, its technical difficulty is not high. Some startup teams and even downstream manufacturers can complete themselves.

This also directly leads to a large number of companies that have no technical content, but a large number of AI -name companies have emerged, making competition in the small application market more intense.

Another embarrassing situation is that, in addition to very few Shang Tang, such as the market that has continued to play the market, it can master AI companies that can grew the initiative to a greater extent. At present, most AI companies are only supporting facilities.

This also makes their development facing a major problem: terminal companies may come out by themselves, neither by technology stuck, but also seize market opportunities. Moreover, the strength of these terminal companies is often unable to confront those entrepreneurial AI companies.

The Cambrian is such a typical.

It rely on cooperating with Huawei mobile phones and smart chips to become famous, but this "honeymoon period" lasted only two years. At the end of 2018, Huawei gave up the Cambrian and opened its own AI chip research and development path.

At that time, the Cambrian not only lost Huawei's orders, but also became a competitor with Huawei, which caused the market to doubt its commercial road.

Major AI companies naturally know the importance of expanding application scenarios, but the expansion of this scene is a systematic project. Many things do not take the unilateral efforts of AI companies.

Even the technical houses of AI companies have insufficient understanding of related scenarios, and they will do a lot of useless work.

For example, in the early years, in order to expand the business to the facial recognition system of the bank in the early years, its team racked the brain to retreat for many days before writing a more satisfactory PPT solution to participate in the bidding.

As a result, the bank took PPT and said: No suppliers have ever wrote a dozen pages of solutions for them, at least 300 pages.

Although the bank is suspected of formalism, it does not rule out the technical houses. It does think of things too simple.

Another problem is that AI is still a new thing, and each application party has the specific circumstances of each family. This has led to the expansion of its application scenarios, often more customized services, which is difficult to standardize and large -scale starting volume.

For example, the Youtu AI laboratory owned by Tencent, in order to complete the smart manufacturing project of a Shanghai factory, not only sent a large number of personnel to settle in the factory, but also consumed the research and development for more than 300 days to really build a set of AI quality inspection for the factory for the factory. solution.

Once a factory is changed, the plan must be re -developed according to different needs.

This further increased the difficulty of broadening the application scenarios by AI, making it difficult to expand its revenue rapidly.

Another constraint is that the utility and application scenarios of AI are positive. The smarter the AI, the more scenarios and data you need to calculate and learn. At the moment, the acquisition of this data is not easy.

For example, in the field of AI medical care, the most difficult thing is the establishment of databases.

Affected by regional and scale, the data quality of each hospital is uneven. AI medical companies must reach cooperation with sufficient hospitals in order to learn from more common data, and then make their own solutions more targeted. And precise.

But because of limited strength, startups can only cooperate with one or several hospitals, which makes it difficult for data to ensure accuracy and comprehensiveness. This leads to a bigger limit for a dilemma: the more unsightly your AI solution, the more difficult it is to find more collaborators; the more you can't find more partners, the more unsightly your AI solution will ... Em

Not only startups, but not just China, including the United States giants, but also headaches in this regard.

For example, the Watson system of IBM has faced doubts of data accuracy.

▲ IBM Corporation New York Genome Research Center Source: IBM China

The questioners believe that the training of Watson system not only uses the imaginary data of false patients, but also the data volume is obviously insufficient. For example, among the eight types of cancer, only 635 were trained data, while the lowest ovarian cancer was only 106 cases.

Including its child -age intelligent auxiliary diagnostic software, it is also considered to be defective in the comprehensiveness of data. The data of the system is basically height data in the southern region and is not suitable for other areas that are inconsistent with the average height of the south.

To sum up, today's AI, it is not easy to be sent in handy, those who can come in handy, but the technology is easy to be replaced. And lack of competition are also lacking opportunities.

So and so on, it is difficult for the AI ​​industry to achieve business and performance in a short time.

【After the cold winter】

Whether it is China or the United States, the AI ​​industry in the early stage of development is still too narrow.

This is the current challenge of AI, and it is making AI companies say goodbye to the highlights that were sought after.

According to iResearch data, in China in 2018, AI's computer vision is only 27.3 billion yuan. However, in the following years, the financing quota has decreased year by year, and there is no such grand occasion.

Including the sluggish stock price of the aforementioned company, it is also related to the challenges of AI.

But the current challenge is also the space for the future.

Even if the road may be twisted, the AI ​​itself is the general trend, and the prospect must be bright. Moreover, although more breakthroughs are facing many difficulties, it has shown the momentum of the fire.

Take the combination of AI in industrial enterprises as an example. Ai Media Consultation Data shows that at present, 80 % of the enterprises have expressed their high attention to artificial intelligence. Nearly 60 % of companies have stated that they will focus on artificial intelligence in the future.

Intelligent manufacturing, smart cities, smart agriculture ... AI will eventually penetrate every place of production. Smart home, smart community, intelligent driving ... in the field of work and life, AI will eventually be everywhere.

Moreover, AI has created brand new value.

For example, Shangtang Technology brings the convenience in the fields of smart transportation, smart business and other fields. For example, the quality inspection plan designed by Tencent Youtu AI is not only more efficient, but also can save tens of millions of labor costs a year;

For example, the emergence of AI in the eyes can provide more accurate diagnosis results for non -ophthalmologists when treating chronic diseases;

Even if we just brush short videos, searches, and family audio, AI has played a vital role in it.

But this is the eve of AI who really became an industry and bred greater business opportunities. A large number of emerging technologies still failed to perfectly combine with reality.

The companies in the between them are also like the Internet in the 2000s, and they have just embarked on a journey to change the world.

Before 2000, the Internet industry only needed a business plan, there was a "E-" in front, and a ".com" later, which could take out tens of millions of dollars from investors.

Investors that year used to describe people's optimistic about the Internet industry: the bolder the entrepreneurial ideas, the more they could recruit dollars.

Until March 2000, "Baron Weekly" gave such a conclusion: After investigating 207 Internet companies, they found that 71%of companies were losing money. 51 companies' cash will be used up within 12 months. This includes Amazon.

Everyone in the future also knows that the Internet bubble breaks down and the stock price continues to fall.

But the fact that the Internet has developed to the imagination of anyone who has exceeded 2000 years, including a large company that has been born beyond anyone at that time, and even a new leader in the development of the world economy.

If you have to say that today's AI foam is broken, it is probably equivalent to the breakthrough of the Internet foam in 2000.

The difference is that today's AI is actually much more than the Internet at the time.

For example, Shangtang Technology not only has real revenue growth and business expansion, but also has 1,1494 global patent assets as of the end of 2021, 78%of which are the technical foundations of invention patents.

And Yuncong Technology, also one of the AI ​​Four Little Dragons, its operating system focuses on occupying a pan -intelligent entrance. You can serve all devices and serve all kinds of software application manufacturers. Under the economic effects of scale, Yun's technology profitability continues to improve. Yun's comprehensive gross profit margin increased from 21.7%in 2018 to 37.01%in 2021.

And a group of enterprises including Shang Tang and Yun from it may truly build China as a highland of AI development.

In early October 2021, Nicolas Chaillan, the first chief software officer of the United States Air Force, said that China has led the United States in artificial intelligence, machine learning, and network technology and is moving towards the world.

Salon gave a pessimistic judgment: "We have not enough competitiveness to compete in China in the next 15 to 20 years. This is a matter of nailing on the board."

Sharang did not say the reason: the most applied scenario and population in the world will not only practice the smartest AI, but also the largest AI market for a long time.

Once the market is really flourishing, AI companies that survive the winter may jump out of the next Tencent and Ali.

——End —————

Welcome to pay attention to [Huashang Tao strategies], know the characters of the wind, and read the legend of Tao Tao.

Copyright, prohibit private reprint!

Part of the picture comes from the Internet

If it involves infringement, please contact to delete

- END -