Reprinted article: First overcome the panic of computing power demand, and then reach the intelligent future created by "digital construction"丨See 2033
This article is reproduced from 36氪
"Seeing 2033" Issue 7: Feng Dagang x Wang Xiang x Zhuang Binghan
Facing the next 10 years, we must have the confidence to face the uncertainties of the future, and the courage to move forward pragmatically in the face of the storm.
Back in 1997, ZTE, which had been established for 12 years, officially listed on the Shenzhen Stock Exchange’s A-shares, accelerating its growth in a more fierce market. Also this year, 27-year-old Wang Xiang joined ZTE and rushed to the "new battlefield" together. In the blink of an eye, Wang Xiang and ZTE have been together for 26 years.
Over the past two decades, ZTE has successfully reshaped the ICT market landscape and become another giant that can compete with giants. For Wang Xiang, the most fulfilling thing is redefining market rules in an era of rapid development of network intelligence.
From the 3G era to the 5G era, every trend breaks out, and every round of technological competition from the Internet of Things to artificial intelligence, Wang Xiang sees the direction of large-scale application of the next generation network of NGN, and sees the future world of interconnected, intelligent, virtual and Trends such as technology evolution and cloudification.
As the senior vice president and chief strategy officer of ZTE, what Wang Xiang has to do is to turn what may happen in the unpredictable future into something that is sure to come step by step.
This is destined to be a road paved with thorns, but ZTE is not going it alone. Relying on the huge ecosystem formed by years of international strategic layout, ZTE already has a group of partners working side by side. For example, its strong alliance with the global semiconductor giant Intel will enable ZTE to continue to become a powerful player in the 5G and AI era, and even in the next 10 years. An important source of support for leaders.
For Intel, in the nearly 40 years since it has taken root in the Chinese market, Intel China has become one of the company's most important business areas. Especially in the digital economy era where computing power has become a new productivity, this giant has also entered the Intel China 2.0 era, which means that Intel will officially move from "Intel in China" to "Intel with China".
This is an evolution and adjustment involving the company's overall strategy, also known as Intel IDM2.0. As the name suggests, Intel in the 2.0 era will build an industrial ecosystem in a more open and inclusive manner, using its expanding third-party foundry and product footprint, as well as its global internal factory network, to serve the Chinese and even global markets.
Because what Intel sees is the digital future supported by semiconductors, Moore's Law that will still be effective 10 years later, and the infinite possibilities in the periodic table of elements.
Looking into the distance during the voyage and grasping the direction of the tide are the missions and tasks of corporate leaders. This is also the original intention of 36Kr’s exclusive interview column "Seeing 2033": to find those who have changed the past and are therefore most capable of changing the future. People were asked about their views on the business society in the "next 10 years" and why they were willing to devote 10 years to one thing.
We set the dimension at "10 years" instead of this year and next year, and we don't talk about specific company businesses but "others" because we believe in the famous motto in the investment circle: people tend to overestimate the future of the next one or two years. changes and underestimate the changes in the next decade.
In a three-way dialogue with 36 Krypton, Zhuang Binghan, vice president of Intel Marketing Group and general manager of data center sales in China, saw the sustainable future that intelligence brings to social enterprises. At the same time, he also calmly examined the The popularity of ChatGPT caused panic in the computing power market demand.
What Wang Xiang of ZTE sees is a future where the digital industry is rapidly maturing and popularizing, the value of the industry is increasing, and the business value will even be greater than in the Internet era. In this future, underlying technological innovation and ecosystem construction will become more important strategic goals for ZTE, and thinking about green energy cannot be underestimated.
Despite the ups and downs in the business world, opportunities and difficulties are always present. But for Wang Xiang, the biggest enemy ZTE faces in the AI era is not from the outside, but whether the inside can adapt to this change. Fortunately, resilience is ZTE’s most proud quality.
The following is an excerpt of the conversation between 36Kr and ZTE Wang Xiang and Intel Zhuang Binghan:
About 80% of enterprises will adopt smart factories in the future
36Kr: What changes do you think will happen to smart factories in 10 years?
ZTE Wang Xiang: Within 10 years, the digitalization of the industry will be greatly popularized. Basically, according to our judgment, about 80% of enterprises will adopt smart factories in the mature stage of development.
36Kr: How will the standards for smart factories differ between today and 2033?
Wang Xiang of ZTE: Under current standards, an assembly line does not require many people to operate automatically, and it can be defined as a smart factory.
36Kr: In addition to the decrease in the number of people, are there any other changes?
Wang Xiang of ZTE Corporation: People are part of it, and we are more concerned about quality and efficiency. To a certain extent, in addition to efficiency, the quality after the introduction of intelligence will be significantly improved than before.
Intel Zhuang Binghan: Cost reduction and efficiency improvement can be achieved through various types of intelligence, and even sustainability can be achieved. Including one of Intel's social responsibility goals, it is proposed that by 2030 we will reduce carbon emissions and achieve the renewable utilization of more than 60% of waste. This is not only to save labor costs.
36Kr: Is it possible that in 10 years there will be almost no people in the factory and the workers will be working thousands of miles away?
Intel Zhuang Binghan: This is definitely possible. Just like when we hold meetings now, people do not have to be on site. In the future, many activities that require hands-on (activities) and feet (activities) do not necessarily need to be on-site. For example, in autonomous driving, you can imagine that the driver can drive remotely in front of a very high-definition screen; or a doctor may not be able to perform a very precise surgical operation at the scene. In fact, many of the things we are talking about are already happening.
36Kr: In fact, there is a saying that the future is already here, but it is unevenly distributed. In other words, much of what we are discussing today that will happen by 2033 may have been completed today.
ZTE Wang Xiang: Yes, some things have been realized now, but because it may be more costly or technically difficult to implement (and other factors), the future is actually a process of transforming these technologies into industry.
36Kr: Today, the popularity of smart factories is not high. What do you think is the main obstacle?
ZTE Wang Xiang: Smart factories are a combination of cross-domain technologies. The first point it involves is that ICT (information and communication technology) has to enter an unfamiliar industry and combine unfamiliar technologies to solve problems, which will face certain technical barriers; the second point is that there are differences in technology integration and business processes in different industries. If there is a certain relationship, then it needs to be combined with new technologies in the process, which is another relatively big challenge. At the same time, if we subdivide a technical point, we call it "Gantongsuan Intelligent Control Security". First of all, "sense" refers to seeing and knowing what kind of sensing device it is, "passing" means that what the "eyes" see must be transmitted to the brain, "calculation" refers to whether the CPU (brain) of the device is fast, and "wisdom" means The artificial intelligence algorithm of the device, "control" is how the device controls the robotic arm after sensing these. Another one is "security", because it involves data protection and on-site execution security. The combination of these 6 can solve one point of the problem. We at ZTE solve about 40 points of automation every year. Our smart factory took almost three years to solve and now solves about 120 points of automation, such as circuit board manufacturing. I think that in about five years, almost 200 points will be able to cover this field. But if you change another industry, such as cement manufacturing, then it may have to solve another 200 points, and it will take a certain amount of time. So this cannot be popularized quickly in the industry.
36Kr: Can this speed get faster and faster?
ZTE Wang Xiang: This is for sure, because when the current development matures and is gradually productized and standardized, other places can be combined . So now we have a platform called Digital Nebula. What problems can it solve? That is, every time I automate a point, I split the experience of this point into elements. When I solve the next point, I can combine the elements to improve efficiency. In the early days, due to the lack of elements, we didn’t know how to perceive this environment, how to transmit it, and what to use for calculation. These were all early confusions. But the further we go, the higher the similarity will be. For example, if cameras are used in this place, cameras may also be used in many places. Then next time we face the same scene, we can directly put the corresponding images on the Digital Nebula platform. Drag the sensing module over, and when the communication is to be solved in the second step, drag the communication module over, so that the line is completed.
36Kr: How does Mr. Han view this issue?
Intel Zhuang Binghan: I think the first thing is whether the purpose of cost reduction and efficiency improvement has been achieved, and the other is the sequence. What does that mean? That is to say, it is actually very difficult to renovate an old factory. If an old factory needs to be shut down for renovation, then personnel training must be conducted, which will cause great interference to the factory's original production business. It would be better to import these things when building a new factory. Technology is sometimes not the most important factor.
ChatGPT lit
Panic demand in computing power market
36Kr: What impact does the AI industry have on chip technology architecture?
Intel Zhuang Binghan: Everyone has always thought that Intel is a CPU company. Now we have also launched XPU accelerators for various scenarios, including AI accelerators and FPGAs, which can solve more problems in specific scenarios. Now the industry has turned to the concept of packaging, which is to package different technologies on one chip. The corresponding process also needs to be improved. Through improvements, we can make the transistors smaller and consume less power. We are now also looking at neuromorphic chips, quantum chips, etc. We hope that the black technology we are discussing can be mass-produced in 10 years.
36 Krypton: There is a saying that the metaverse developed by Facebook has not really been implemented to this day, and a very important reason is the computing power issue. If we get to 2033, what level of computing power will be needed to support our vision?
中兴通讯王翔: Facebook的元宇宙涵盖范围很广,它所谓的6大核心技术目前都没有达到理想状态,仍是短板,很难呈现出一个完美的画面。 从目前来看,最显著的短板不是算力,因为算力靠现在的技术似乎也能达到。按照摩尔定律,10年后算力差不多会提升100倍,我觉得这个速度是能够满足我们对元宇宙的一个需求。
36Kr: Some time ago, everyone was discussing the problem of insufficient computing power of ChatGPT, because it was impossible to buy chips. What do you think?
Intel Zhuang Binghan: After ChatGPT came, everyone discovered that AI could be used in this way, and it suddenly became popular. In fact, it also created a stage of panic in demand. What does demand panic mean? That means it doesn’t matter if it works for me or not, but if I don’t, I lose. For example, the original market demand was only 100, but due to this panic, the market demand suddenly increased to 300 or 400. In fact, from a supply perspective, I think as long as there is market demand, supply can definitely keep up. Including Intel, now we have launched the concept of IDM2.0, which means that we can not only provide CPU chips to customers, but we will also expand our production capacity and help customers manufacture chips. Therefore, we are expanding production all over the world, and our production capacity can always keep up.
36Kr: Within 10 years, is it possible that chip demand from other industries such as autonomous driving or smart homes will exceed today’s Internet industry?
Intel Zhuang Binghan: Because of power consumption, you can never put a very high-end computing power on a terminal product, including cars and mobile phones. If you want to respond to certain applications very quickly, you can rely on the computing power of your phone or car. But for large-scale model training, real-time training, and push after training must be placed in the cloud or edge.
36 Krypton: A large amount of computing power is actually unevenly distributed, such as the problem of "calculating in the east and calculating in the west". What do you think?
ZTE Wang Xiang: The logic of China’s “East to West” proposal is similar to the previous “South to North Water Diversion”. The rationality of this logic lies in the fact that we have vast land in the west and north, and the second is that there are relatively low temperatures. , and the third is to have broader energy support, so that through global coordination, it can optimize the global cost.
36Kr: Do other countries have a similar practice of “counting in the east and calculating in the west”?
ZTE Wang Xiang: I haven’t seen it yet. Other countries, especially geographically large countries, actually have a more balanced population distribution compared to China, while China has an unbalanced distribution of population and resources. This feature is rare in the world. Of course, our proposed “numbering in the east and counting in the west” also faces new difficulties. Because data generation and data storage are not in the same place, the distance creates a delay problem, which may make you dissatisfied with the results. In principle, everyone now believes that the inflection point of satisfaction is 20 milliseconds. If the data from point a to point b is longer than 20 milliseconds, the business may not be acceptable.
36 Krypton: Human-assisted autonomous driving, where the driver operates in the cloud, can this be achieved through "numbering from the east to the west"?
ZTE Wang Xiang: All artificial intelligence is training and reasoning. I think the training of autonomous driving must be placed in the cloud. It does not have to be real-time, but for inference, decision-making must be done locally. "Counting from the East to the West" solves most of the difficulties in our scenarios, but for special scenarios, solutions that meet the service needs of the scenario must be used. The same is true for the distance we just mentioned. Nowadays, many distance delays are caused by forwarding. Each time it is forwarded, it may be delayed by a few milliseconds. The more times it is forwarded, the longer the time will naturally increase.
For ZTE, we are wondering whether it can be achieved through some short forwarding, such as launching a long-distance powerless relay transmission solution, which can meet the transmission requirements from the west to the east, and even to Guangdong, Hong Kong and Macao, and at the same time, the delay can be reduced to 20 Within milliseconds. This involves the reconstruction and design of the overall network.
general artificial intelligence
Hard to beat human industry know-how
36Kr: Will the concentration of resources in the AI era increase? Will it be a monopoly resource?
Intel Zhuang Binghan: I think China has made a good decision, which is to treat computing power as a livelihood resource like water and electricity. This direction is to ensure that computing power will not be monopolized by certain people or companies. In fact, "numbering in the east and calculating in the west" is also consistent with this concept, because through the panoramic layout of the country's computing power network, computing power can be distributed fairly and reasonably to everyone who needs it.
36Kr: In the AI era, will the concentration of the entire business become higher or lower?
ZTE Wang Xiang: The industry must be fragmented because the differences are huge. But for ordinary consumers, I think it will gather, because for AI, the more data it has, the better its reasoning will be. When there is too much data to a certain extent, the Lanchester effect will occur, and its results will be better. So we believe that in the consumer field, the future will be a concentrated scene. However, because the knowledge of each industry is very different, it is difficult to use one platform to cover the detailed knowledge of all industries.
36Kr: Will general artificial intelligence defeat our industry’s know-how?
ZTE Wang Xiang: There is corresponding data in the industry, but it is difficult for you to fully obtain the know-how in it. When all know-how is known, maybe general artificial intelligence will be stronger, but I think it will be difficult to get to this point.
36Kr: Will the knowledge of vertical segmentation of each industry be open to large models?
ZTE Wang Xiang: The first is the openness of this knowledge. I think every company will protect its core intellectual property rights. A dictionary is just an introduction to a professional field, but for ordinary artificial intelligence, it is called professional. There is a saying that "Don't use your hobbies to challenge a person's career." So I think it is difficult for artificial intelligence to obtain this knowledge in the profession, and there are certain barriers to obtaining this knowledge.
36Kr: How to protect users’ data security in the AI era?
中兴通讯王翔: 数据越多,责任越大。 人工智能可能带来最显著的风险就是隐私安全等,这需要在技术上逐步解决。我们自己也做了很多工作,从数据的采集、传送、使用、分享,包括未来有些数据要注销,这些全过程我们都要做相关的工作,来保护客户数据的安全。
36Kr: ZTE’s layout in the emerging digital industry?
ZTE Wang Xiang: We focus more on the bottom layer of the equipment and the ecosystem. The bottom layer is our core technology, such as communication chips, and basic software such as operating systems and databases. In terms of ecology, competition in the future will become more and more complex, and it will be difficult for a single company to survive in the industry, so we need to develop more ecological partners. For example, we and Intel are strategic partners in terms of computing power. When we are doing digitalization in different industries, we must also work with industry leaders, otherwise it will be difficult to penetrate industry knowledge barriers.
36Kr: What is there to worry about in the bright future?
ZTE Wang Xiang: Energy. ZTE submitted its commitment to science-based carbon targets in May this year. It took us two years to figure out all internal carbon consumption paths, and we also used a lot of technical means to make changes.
artificial intelligence
“To Be or Not to Be”
36Kr: Some time ago, Musk and others jointly signed an open letter calling for a moratorium on the development of AI. Would the two of them sign it if they represented individuals?
ZTE Wang Xiang: Improving machine intelligence must be the future direction. But just like a movie I once watched, it often asked what the machine was born for, and the final answer was "born to be a human." Since we are born as humans, we must comply with human nature and human rules.
In fact, at present, we are mainly worried about AI abusing intelligence in the future. When intelligence is abused by machines, humans may not be able to compete with it. We can calmly think about our future positioning and the relationship between AI and people. When it becomes more and more powerful, what rules should it comply with and what kind of humanity should be injected into it.
36Kr: In your personal opinion, do you think AI is a human being or not?
ZTE Wang Xiang: I think we can think about it calmly, but we must continue to move forward in the future.
Intel Zhuang Binghan: It's like opening a Pandora's box, it's hard to put it back again. AI will definitely bring many things that benefit the world to mankind, but it may also have some hidden dangers. Many things are not one-sided, and the overall situation must be considered. But I feel that this train has started and it is difficult to stop it now, and it may not be necessary to stop it.
36Kr: So it’s impossible for AI to become a human, right?
Intel Zhuang Binghan: At least the AI we have seen so far, we think it still has no emotional ability. Can it be quickly improved so that it can have emotions and thinking capabilities, and can judge and decide what is right and wrong by itself? At this stage, we have not seen such AI, but it may be possible in the future.
Wang Xiang of ZTE: We cannot be afraid of a good technology just because it has certain risks. In fact, it is not the technology that is bad, but who uses it. Everything has risks, and we need to see which one is greater, its contribution to human society compared with its risks. When it brings greater benefits, then our main task is to control risks.
Create greater business value than the Internet
36Kr: Do you think there will be a virtual world that restores the real world 1:1 in 10 years? Just like the "mirror world" mentioned in Kevin Kelly's book.
Wang Xiang of ZTE Corporation: The mirror world is a projection of life, which extends people’s physical environment. I think it will happen in the future, just sooner or later.
Intel Zhuang Binghan: The mirror world is not created overnight. If it evolves into the category of education and entertainment and allows more people to participate, then I think it will not be far from the mirror world just mentioned.
36Kr: How do you view the difference between the mirror world and the metaverse?
ZTE Wang Xiang: When we talk about the mirror world, we are more concerned about the connection between the digital world and the natural world. But when we talk about the metaverse, what we care about is when I integrate into a digital world, what is my lifestyle? Can I generate transactions with it? What is the connection between these transactions and my life? I don’t think we need to distinguish between the mirror world and the metaverse.
36Kr: Would you like to live in a world similar to "Ready Player One"?
Intel Zhuang Binghan: I am very willing to try and immerse myself in this world, and I hope this can change my life and make my life more interesting and efficient. And I also hope that more companies involved in such applications will participate, so that the Metaverse or the Mirror World can become richer, not just for certain applications.
36Kr: Intel has been one of the most successful multinational companies in China over the past 38 years. Do you think it will achieve greater glory in the next 10 to 20 years?
Intel Zhuang Binghan: We have seen that the Chinese market has brought many very high-level innovations. For example, our partner ZTE has many innovative ideas in various fields. We are also very honored and hope to actively work with our Chinese partners, not only to develop Chinese industries, but also to serve the world.
36Kr: Assume that the next 20 years will be the climax of China’s digital construction, just like the past 20 years have been the climax of China’s Internet industry. Do you think that in the process of digitalization, is it possible to create greater commercial value than the Internet?
ZTE Wang Xiang: It will definitely bring greater value. In the early stage, the Internet mainly benefited people and provided convenience by guiding people's consumption. But the next step in shaping the industry is more about creating new industrial value. Only when the new industrial value increases will circulation become more valuable.
In fact, in the past, the Internet increased circulation efficiency, but in the future we will increase basic value. When the basic value increases and the circulation efficiency becomes higher, the overall market prospects and space of these two industries will also become greater.
This article only represents the media’s views
The copyrights of images and other materials in this article belong to their respective owners
Want to see more "core" information
Tell us
with your
likes
and
watching
~
© Intel Corporation. Intel, the Intel logo, and other Intel marks are trademarks of Intel Corporation or its affiliates. Other names and brands mentioned in this article are the property of their respective owners.
Featured Posts