Capital.
Several phenomena have been found in the current primary market:
1.Basically, they are all professors and experts from universities and research institutes, which is equivalent to the resignation of senior executives of Internet start-up companies to start businesses;
2.No mention of income model, no mention of financial indicators and expectations, only mention how much money is needed;
3.It mainly talks about the technical principles and the grand industrial problems to be solved, but does not mention who the customer is, what are the problems, and why do you pay?
4.The product is mainly in the prototype and pilot stage, with a basic financing amount of more than 100 million yuan or a valuation of more than 100 million;
5.Investors do not ask business questions when interacting with them, mainly with the entrepreneurs in the most technical details.
Snowball Shen Junbao.
I still say that if the mourning stock does not carry out a level of institutional reform similar to the "equity division reform" back then, according to the current "pseudo registration system" blood drawing model, how much money GJD invests is a drop in the bucket.
Even now, IPOs are not suspended: it shows that the real bottom line of regulation has not yet been reached.
He Xi. Healthcare.
At present, under the Healthy China strategy, the state strongly promotes early diagnosis and early treatment in the fields of cancer, cardiovascular and brain health, but the price of physical examination items in these fields is relatively high, and the equipment to be used for physical examination in these fields is also relatively expensive, such as CT, MRI and gastrointestinal endoscopy, etc., which cannot be covered by 500 yuan, which has formed the conclusion that "1000 yuan cannot do effective physical examination".
Fang Yixin, founder of Ruici Medical.
Consultation.
The marketing consulting company has to fight to the death with the legal department of the client company, and suggests that the boss change the evaluation mechanism of the legal department. Even more terrible than the audit is the internal audit, because the department in charge of the internal audit will make the standards ten times stricter, not for the benefit of the company, but for their own sake.
The legal department is not to prevent the brand marketing department from "playing the sidelines", but to keep it 100 kilometers away from the border. Within an enterprise, the business department is often in a weak position in front of the legal department, which requires external intervention.
We believe that a legal counsel who is only responsible for saying no on the principle of "zero risk" is enough to give a monthly salary of 2,000 yuan, because his guidance is that there is a risk of being hit by a car when going out, and he is not allowed to go out. Such work is not only untechnical, but also destructive, lazy, and obstructive. And a legal counsel who is truly responsible and provides compliance advice, risk assessment and solutions can give an annual salary of more than one million.
Hua Shan, chairman of Shanghai H&H Marketing Consulting.
aigc
For the technological development trend in 2024, Fu Sheng gave four predictions:
First, in 2024, more than half of the 100 billion model will wither, and the privatization of the 10 billion model will bloom. Second, the opportunity to surpass Openal comes from large-scale model application companies. Third, the "digital boss" will become the standard configuration of enterprises, and the board of directors will evaluate the time of use of AI by CEOs. Fourth, embodied intelligence is very popular, but humanoid robots must not be mass-produced.
Fu Sheng, Chairman and CEO of Cheetah Mobile.
There are only two opportunities for AIGC in China: 1) reducing costs and increasing efficiency in existing scenarios; 2) The technology is very difficult, and the United States can't do it, so China relies on some manual and some AI to directly deliver the results to compete in sales efficiency and delivery efficiency. Instead, all AI will soon become a red ocean. Fu Sheng's three diagrams are particularly good, many vertical scenarios are optimized with open source small models The results are far better than those of general large models, and the importance of vertical scenes and data far exceeds the large model itself.
Zhu Yuhu, Managing Director of GSR Ventures.
A few of my points about generative AI and large models** (guesses).
1.The large model is the operating system of the AI era.
2.Like the operating system, there will be only a few mainstream large models left.
3.Similar to operating systems, open source and closed source models will coexist, there are 1-2 closed source models, 1-2 open source models, and open source models will derive different branches and versions like Linux.
4.As in the Internet era, 2-3 new AI giants will be born, mainly at the AI application level.
5.Countless AI startups live and die.
Dr. Yuan Chen of Silicon Valley.
Knowing that everyone wants to hear about the big model, I will try to use the most objective facts and opinions to say it.
Of course, it may not be completely correct, so let's make an investment reference.
After the large-scale model fever in 2023, in fact, there has been a cooling phenomenon in the second half of the year, and the capital market has begun to go to extremes. The main factor in this is that on the one hand, the industrial development objectively exists, and on the other hand, the characteristics of our Chinese, like to swarm bees. Just like new energy vehicles and photovoltaics and batteries, but also like pig breeding in the past, and so on, see that everyone is coming, no matter what the purpose, no matter what the strength, whether it is related or not, let's mix it up first.
Therefore, it is inevitable to shuffle and clear, and the cost of a computing power cannot stand it, but it only fattens Nvidia. I have complained before, but this is not a phenomenon that can be changed by anyone, so it can only be ironed out with time and the pain of more people. According to incomplete statistics, there are 254 large models with more than 1 billion parameters, and most of them will die in the end, except for some specific vertical categories that can struggle.
Then the general model is divided into four categories, one is open source like Tongyi Qianwen, which can make money on Alibaba Cloud; One is Byte and Tencent, most of which use their own business to try and play, and will mainly do C-end in the future; One is to really solve practical problems, to improve production efficiency, liberation and productivity as the core, similar to iFLYTEK Xinghuo; There is also the fact that I brag about how good I want to do everything, so I won't give an example.
But now I encounter a problem, the C-end application of the large model, it is difficult to make money in the short term, and the development of applications on the basis of the general large model (the one that Robben has been calling for) has one of the biggest bugs is that there is still a gap in our domestic large model, and your own gap and performance can't keep up with others, how can you develop secondary applications based on you? Therefore, in addition to the bytes with deep pockets, Ali, Tencent can cultivate user habits on the C-side, most of the other large models are most worried about landing, of course, there is continuous iteration, but the higher the iteration, the more difficult it is, and the higher the cost of computing power, so the competition for large models in 2024 will be clear, and most of them will naturally fall behind.
The immediate effect is to add a large model to their own business and products to improve performance and market, and secondly, the landing of the GB side will take precedence over the C-side, so the companies in the GB-end business that have been widely criticized by the capital market in the past may have a first-mover advantage during the landing period of the large model, and there is also a huge expectation gap. For example, the joint team of iFLYTEK (technology) and Huawei (chip and computing power) has landed some targets at the end of last year, such as projects such as Lichuan, Hubei and Jinan, Shandong.
Of course, there is another consideration on the GB side in the future, that is, the problem of information and innovation, that is, I have been saying before, whether the computing power is domestic computing power, whether the large model is independent research and development, independent and controllable, and whether the capital behind it, including the shareholders in the secondary market, is mainly in the mainland. Because behind the big model is the data.
Therefore, to sum up: the large model in 2024 will be differentiated or even cleared, because the decline in demand is not very friendly to computing power companies; On the one hand, the domestic large model has to race against time for iteration, there will be a process of catching up with each other, and on the one hand, it is necessary to deal with the cost of investment and research and development, and on the other hand, it is necessary to strive for the competition of landing.
Back to the stock price, it still depends on how the market reacts, if you look at iteration and landing, technology and income, then it is not excluded that there will be good investment returns; If you look at profits as simply as in the past, all large model companies in 2024 will most likely not reflect much profit, so they may continue to wait.
Of course, try for the best and expect the worst.
Brother Mianzhou. Artificial intelligence.
SenseTime's share price fell below HK$1. The market capitalization is more than 30 billion Hong Kong dollars. Although many people will say that enterprises in the pre-AI era have not yet found a sustainable business model, and it is difficult to make money, and they have to rely on non-standard projects, but after all, SenseTime represents a class of high-tech artificial intelligence capabilities in China, and failing to prove itself in the capital market is also a loss for the entire industry. Now that everyone is making a big model, it's hard to say whether such a story will be repeated again.
Xiang Oda.
There are relatively few opportunities for domestic artificial intelligence to participate, and there are certain restrictions in various fields.
For example, infrastructure, servers, optical modules, etc. cannot get key things, and they cannot be exported.
For example, the development of large models of a search company was okay, but now it is also restricted.
For example, the more critical AI needs to be trained with data, and this data must have global science and society...
For example, there are signs of transfer in advanced packaging....
The one that can currently participate is HBM:
1.HBM, this seems to be not stuck, the main Korean semiconductor companies are under financial pressure, and they dare to contradict....
2.There is one in China that can supply HBM packaging and testing equipment, and it has risen.
3.The consignment HBM has also risen, and it is currently high.
Anyway, that's it.
The Grand Lord of London.
AI is the future. However, the corresponding entire chip industry chain bypasses China (except for the non-core of optical modules), and the large model is also, so on, AI is doing its own thing, and electric vehicles are also clearing the field through various **non** barriers. These are all industries that represent the future, so this is a historic moment, reform and opening up is particularly important, and only by integrating into the world can we not be decoupled. It is important to be self-reliant and self-reliant at a high level of science and technology, but the premise is to integrate and develop with the world.
Snowy westerly winds in Zhejiang.
A large part of the reason why the story of AI PCs in China cannot be told is that there are no more everyday applications that support PC and the web. Even if your agent is as strong as a human, you won't be able to do many things on your PC.
Comment on the corpse.
Publish the top 10 tech adoption trends impacting 2024, including:
1."Four-Computing Fusion" of High-Performance Computing;
2.Multimodal agents accelerate the AGI process;
3.AI accelerates the evolution of humanoid robots "hands and brains";
4.AI+ genetic computing to interpret the code of life;
5.The digital interaction engine stimulates super digital scenes;
6.Immersive** gives birth to 3D presence;
7.Brain-computer interface from medical breakthrough to interactive revolution;
8.Satellite-to-ground direct communication promotes ubiquitous network coverage;
9.evtol accelerates the empty ** line to the new era;
10.Multi-energy streams collaboratively reshape virtual power plants in real time.
Tencent Research Institute.
New energy vehicles.
Recently, there are four main Chinese car companies whose brands have appeared on CCTV, and the frequency of their appearances is ......
Xiaomi (technological innovation).
NIO (from a big car country to a strong car country).
BYD (leading the global new energy industry).
VOYAH (exported to Europe).
Notice that I'm talking about "frequency of appearances".
His Highness Tomato.
The question now is whether the intensification of involution by Chinese automakers will do more harm than harm or more harm than good to China's entire auto industry. Although the involution has exercised the king of volumes, it has also fully eliminated many Chinese companies, whether these companies should have grabbed market share in the overseas market, rather than competing in a pot and not making money in the end. The worst situation is that the domestic car is dying of involution, and then it is reaped by foreign capital with the power of overseas markets and capital.
Xiang Oda.
Although the report cards of all walks of life in 2023 have not yet been announced, the big rankings have already been settled. Among them, the most dazzling Xueba is none other than the automobile industry, which can be clearly said to be more outstanding than the breakthrough of the high-end chip industry.
Because although the breakthrough in the high-end chip industry is also critical, the overall situation is still a defensive situation, and the current result is to ensure that they will not be trapped by their opponents. The automotive industry is different, what new forces and old forces are next door to cross-border forces, according to their own understandings, thousands of horses are running together. Looking at the results, it is not only the No. 1 automobile producer for many years, but also the No. 1 automobile exporter immediately.
In terms of total automobile exports, although there is not much gap with the second place, the development direction of our products is more reasonable and more future-oriented, and the time is in me. Not only that, our main battlefield with the second place is Southeast Asia, which is closer to us, and the geographical and climatic conditions are also suitable for the tram to play an advantage. In short, the advantage is mine.
Rhino Xiaoli. Semiconductor chips.
The day before yesterday, the CEO of a domestic MCU said that there are currently more than 400 MCU companies in China. So in this case, even if there is consolidation, even if there are several integrations, it will not work for the industry. After the merger, a few more companies will come out, maybe after the merger of the two, but after 1+1, there will be more than five companies out.
Xinmou researched Gu Wenjun.
China's imports of integrated circuits fell by double digits in both volume and value last year.
There are two main reasons.
1.Last year, the end market was weak and demand was sluggish, especially for chip design companies.
2.U.S. export restrictions on high-end chips to China.
Of course, there are also factors such as the decline of chips.
Xinmou researched Gu Wenjun.
Under the premise of guidance or approval, when each FAB project is established, it is a dislocation competition, and there is little overlap with the existing domestic subjects. But once the project is successful, it means that you have everything and I have everything, and everyone does exactly the same.
Xinmou researched Gu Wenjun.
Finance.
The financial industry has always been a cyclical industry: when the market is good, it is really profitable, making money for three or even ten years a year; When the market is bad, it just doesn't make much money, and the practitioners don't do it well, they lose their jobs.
Therefore, when the market is good, we must work hard to make money and eat until we can survive; Don't worry about making a lot of money, because a considerable part of this money is also the ration for the cold winter in the future.
The most feared thing is that I take it for granted that "the market is good", and I "should" make so much money all the time; Then, according to this "expectation", ...... is consumedThen when the cold winter comes, you will be very miserable.
I came from the Internet, and the Internet company did a project: if it was a new project, the growth would not be as expected; Or it is a mature project that can't make money - cut it directly, and the entire business department will be cut from top to bottom.
In the Internet industry, even a successful product like Tencent and WeChat, it is not possible to say that it will not be subverted in the future.
But the manager of the cemetery will not be worried: the track of subjective bias will not be subverted by the new format that emerged from nowhere, right?
In addition, in the financial industry, there are high-value licenses: even if you lose money to customers and take management fees, the company will not go bankrupt.
That's why I said that finance is really a very good industry, and there is no risk and trouble that the entire business line has been cut off, and the business format has been subverted and the company has gone bankrupt.
Financial practitioners, don't "be in the midst of blessings and don't know blessings".
He Xi. Photovoltaic.
This year's photovoltaic industry is also a reshuffle year, and a large number of companies will be cleared:
1.The fierce ** war will continue for a long time;
2.Phased absolute overcapacity, accelerated clearance;
3.More than 100 billion PV projects have been terminated or postponed;
4.The oligopoly effect is prominent, and the industrial concentration is rising instead of decreasing;
5.The IPO policy tightening cycle has been extended, and some companies are facing greater pressure;
6.Financing is tightened, and money is not so easy to take;
7.Cross-border players are under tremendous pressure, with more than 10 listed companies withdrawing from PV;
8."De-globalization" is highlighted, and the global market expansion is facing new challenges.
He Xi. Internet technology.
When the market value of Pinduoduo surpassed Alibaba, many articles came out to write about how Pinduoduo was successful and how effective the organization and management were, which were all correct, but if you study the changes in the market value of the two, you will find that the market value of Ali is closer to Pinduoduo, which is much greater than the market value of Pinduoduo to Ali. It's not that Pinduoduo is rising fast, but Ali is falling hard. It is normal for outstanding students to fall to the bottom and be overtaken by other students. In fact, it is more worth studying why Jingdong has fallen into a fraction of people's lives.
I have a point of view, that is, if a company is constantly spinning off, then the management's energy is on capital operation, which is not a good thing for a technology company, and the resources to be put on innovation are not enough. I hope Ali doesn't want to go down this path.
Xiang Oda. by A round of financial consolidation January 1).