One, who created the chip miracle?
Is an integrated circuit chip, we use every day, such as integrated circuits among the families have used as many as three hundred. We have some electrical repair in your own home, you can see there are a lot of black boxes, what are these black squares are? What we call integrated circuits and chips.
There are a large number of basic components of integrated circuits, called the transistor, there may be dozens or even ten billion one hundred million. Principle transistor is very simple, but really make such a transistor invented, human or after a very long period of exploration.
We know that the world's first computer was invented in 1945 the University of Pennsylvania, we use the so-called tube, probably about two centimeters in diameter, has a height of five to six centimeters, it will pass on power after shiny, like a bulb-like.
Such a computer with a 17500 tube, a lot, but very poor reliability of the tube, six minutes to burn a once burned out how to do it? We have to replace it.
In other words, when the computer room where some women will go to power off, replace it with a tube, and then reboot. Such a computer use efficiency is very low, so we urgently need to find a substitute for tube components.
In 1947 Bell Labs in the United States, there are three scientists invented new components later we called the transistor, three scientists called Shockley, Bardeen called, there is one called Bu Laitan, the three scientists won the Nobel Prize in Physics in 1956.
After the invention of the transistor, we see it than we know it to be a lot smaller tubes, smaller than a soy, sesame seeds or even like a, like, very reliable, and it reacted fast.
In 1954, Bell Labs transistor with 800 computers set up the world's first transistor, this computer is to use B-52 heavy bombers, which power consumption is only 100 watts, the most important of its computing speed very fast, reaching one million times per second.
Transistor is very good, but we still think that I was not able to be made smaller transistors? why? You have so many transistors, it still has solder, solder joints Weld would later have poor reliability Weld, then we can not find a better reliability things? So back we appeared integrated circuit, that is, today we talk about the chip.
September 12, 1958, by then a young engineer at Texas Instruments, he called Jack Kilby invented the integrated circuit of the theoretical model.
In 1959, then at Fairchild named Bob Noyce of people, but also was the co-founder of Intel, he invented a method for manufacturing we are using today's integrated circuits - carved mask exposure erosion technology.
So we are talking about today in terms of go, in fact, we are using technology is the invention of technology sixty years ago, but we are constantly in size, smaller precision today only, two scientists invented the integrated circuit influence on humans is very great.
In the 42 years after the invention of the integrated circuit, Jack. Kilby was awarded the Nobel Prize in Physics in 2000, Bob is very unfortunate. Noyce that time has passed away, so he did not get the Nobel Prize.
In 1962, when International Business Machines Corporation, which is IBM, began to manufacture integrated circuits computer in the world in 1964 he released a series of six computers, named called IBM360, extremely powerful, complete scientific computing, transaction processing and other all sorts of content.
Then, after a few years, Intel has a young scientist, the scientist named Ted Hoff, he designed the world's first microprocessor, the Intel 4004 called.
This started when the microprocessor was born, life experience is not so tall, is to use a calculator, is a Japanese company to look for Intel, allowing Intel to help design a chip.
So Intel's your life they buckle down, the final design will give the company called SurfControl Japanese companies, do calculator.
In 1981, that is, ten years later, IBM organized a team that went to Florida to develop an impact all over the world to this day, the major products of humanity, that is, personal computers, and later we called called PC. I was using Intel's 8088 microprocessor, in fact, it's very slow, but at the time was very great.
Therefore, the integrated circuit chip and progress, continue to use the government to the private sector from the original application, such as our general application from military to civilian, from the people at home and replaced by a general, conventional market commercial applications.
Chip sector has a famous Moore's Law. Content which is generally as follows:
When the same price, the number of integrated circuit components can accommodate, will be doubled for 18-24 months and will also enhance the performance of about 40% intervals.
Over half a century, the evolution of chip manufacturing technology continues to verify the level of this law, continue to promote the speed continues to drive the rapid development of information technology.
Chip technology development stage to what level? Whether future development will encounter the limits of it? Moore's Law can continue to work? The miracle chip industry can continue for many years?
Second, how amazing chip technology?
Today's chip technology in the end how amazing? It constantly in miniature, constantly shrinking. Reduced to what extent? We have now done 7 nm, estimated next year, the year after to 5 nm.
They say nanometer mean? Nano did not feel. I do not feel, for example, we can imagine how small. We saw our own red blood cells do? Certainly not seen, but we know that we are a drop of blood is red because red blood cells are red, mapping out blood is red.
The diameter of red blood cells be? Erythrocytes diameter is 8 microns, is 8000 nm. According to today's technology, for example, 14-nanometer chip manufacturing process, about 40 nanometers in size, so we can put the transistor 200 in the diameter of a red blood cell.
So you can think of something so sophisticated, precisely because it is so small, so we can put a lot of things integrated in a single chip up.
We will ask a question, if we follow the walk, went to 5 nm, and then went down to 3 nm, can no longer go on it? We believe that a particular technology may come to a certain time, it will stop, but does not represent a new technology does not appear. Two years ago German scientists have invented a molecule called a new stage transistor devices.
The future, we may phone will become smaller and smaller to the point where we are unimaginable today. Of course, this does not mean that little smaller size, the size of mobile phone chips become smaller, functions become more and more.
But any technology has its limits, can not be without limits, and that it has a point of view from the chip which several limit it? Is a physical limit, which size is too small, there is also the power consumption limit. For example, we have a home electric irons, electric iron power density of 5 watts per square centimeter. 5 watts is very small, but very hot, we definitely not good to go directly touch it.
But the integrated circuit chip it? Generally chips are tens of watts per square centimeter, so often we see on the chip back to a radiator, there are even a fan.
When we reach a power density of 100 watts per square centimeter, the wind has to die, to be replaced by a water-cooled. Among supercomputers to water, cold water here go out there it becomes warm.
Such a thermal power consumption, the thermal effect is very, very powerful, if not controlled, by the year 2005, we have reached the temperature of the chip temperature nuclear reactors, to 2010, when probably already reached the surface of the sun temperature, then so hot stuff might use it? Impossible to use.
So people think of a way, we find a way to put the power down, the original single-core dual-core becomes.
Later extended to mobile phones, there have been a particularly interesting phenomenon, when you buy a mobile phone salesman to tell you to buy this phone now, this phone is 4 core, 4-core powerful, better than that.
Another person told you, do not buy that 4-core, 8-core I've got a good eight-core to 4 cores. what do you mean? In fact they do not understand this question, because we do not monocytes, we put it into a dual-core, made 4-core, 8-core.
From programmability, the single-core is best, but if you want to reach the power of four cores to run, then, should be done single-core power consumption is very high, too hot.
Thermal how to do it? I had to take it apart, in fact, is the complexity of the system at the expense of power to solve our problems. Therefore, the power consumption as a constraint to our development of very important problems.
The second is the difficulty of the process is very, very large.
Among integrated circuit fabrication process, the mask layers it is in fact constantly changing, from the layer 65 40 nm to 7 nm when the layer 85.
So many layers, each run of the day, then, to 80 days to finish, right? So we are now manufacturing the chip takes a long, long time, not the short term can be made, in case there is a mishap, this chip might be scrapped off, so it's the complexity of the process is very high.
The third is that we have seen its high design complexity.
It is precisely because there are so many transistors on a chip, it becomes more and more versatility poor, hence the so-called called "high-end general chips", go in search of a more general solution, then put software imported.
So I often say a word: "Both chips, software inseparable, no ghosts chip software is no software chip is a zombie." We often teaching them good, work which both mean both should organically.
Of course, all these process issues, and that these are technical issues, the most important thing is the economy. Among Moore's 50 years of development, integrated circuit about 55 years it is in the process of price cuts, direct effectiveness of our electronic products are very cheap, cheaper to what extent? Many of our young people every six months for a cell phone, and now we can not change, because what is it? Mobile phones become more expensive again.
The reason is the development of chips due to the increases, the complexity of the investment, its cost is in fact a slow increase of 28 nanometers before our costs are continually on the decline, after the 28-nanometer our costs gradually rise.
So we can predict is the future of our electronic products as no longer continue to cut prices as in previous years, is expected to further price increases, of course, is slowly rising. So we say, the development of chip technology to date, we still do not see its end.
Moore's Law is not coming to an end? This debate has always existed. One thing that was in 1997, in 1998, when a man published an article about, say chips, Moore's Law is dead, gone in one place. He said you see copper interconnects, we had all aluminum, copper interconnects are now engaged in for so many years are not going to engage in, the game is, copper definitely go elsewhere.
Said second chip drain after this thing growing in thin, unable to control, so do the minimum chip 50 nanometers also go elsewhere. He also cited an example, since said lithography using a 193 nm wavelength light, the wavelength we know it will to a certain extent after diffraction becomes false, so finished Moore's Law.
We later learned, we use technology to solve the problem of plating copper; K- so-called high metal gate technology to solve the problem of the so-called media; then use a particular special way; we shot turn on the water, the use of water refraction about shortening the wavelength. So now the lithography machine can not only use our 14 nanometer today, you can also use the 5 nm.
The three full technical breakthroughs, we asked, who is so not opening the eyes? So how Moore's Law? Turns out, is not Gordon Moore himself he said. We see such a scientist, such an important person, he was talking about himself when he may not be able to speak clearly.
Continued development of chip technology continue to make breakthroughs led chip industry. 2018, the output value of the global chip market up to $ 468.8 billion, China is not only one of the most important consumer markets worldwide chip, while also making every effort, marching to the first echelon of the global chip industry. Our chip industry in the end in what stage of development? Catch-up process, which we are facing serious challenges?
Third, who is the biggest buyer of the global chip market?
We have 2014 as a node, then, in 2020, six years, computers will grow 46%, 81% growth in mobile phones, and consumer electronics also rose 48%. So, the growth of electronic products is more and more, faster and faster.
My personal judgment, in our lifetime, if you can not find something can replace semiconductors, electronic products will now probably continue to go this way.
We will always bring electronics products to enjoy a variety of convenient, but the fundamental factors behind it is to break through chip technology. Because there is such strong demand, the development of the global chip industry is very fast.
That is how the semiconductor market distribution? We see the red is China, the bottom of the American market is purple, blue is the European market, the Japanese market is gray, green top is in addition to Japan, China and other Asian markets.
This figure a little alarming, because the Chinese market accounted for 34% of the global market, to $ 158.4 billion, more than a third, which refers to the Chinese market used.
Meanwhile, the 2018 China is the fastest growing semiconductor market, China's semiconductor market grew 20.5%. I think we can imagine how much to buy Chinese integrated circuit, a lot!
4
Demand, insufficient supply of the chip industry, how to force?
We might think, chip industry development of our country does not seem so good, I think we have such feelings are normal.
For example, two years ago, we have a lot to say bigger, I used a word called "scare body in urine." I will give you read a few paragraphs, very interesting phenomenon, he said, "such and such a chip by leaps and bounds, why Americans are afraid of?" And said, "what we stand to what the top of the world", there is a saying " we came back from the United States certain elderly Americans panicked. "
When we met last year, when something, attitude is 180-degree turn in favor of their own "scare urine", but also for everyone to read a few paragraphs, for example: "You do not know how bad Chinese core, you have only read after this article you will know how bad it is. " "China chip in the end how, with a people than I was completely disappointed." I do not know how we are so fragile, confidence in myself did not it? Two years ago the kind of grand ambitions and where to go?
The development of chips it has its objective laws, neither as we imagine so good, nor so bad as we imagine, of course, we still can not meet the demand, but as long as perseverance to go on, our development will be able to go on we want that level up.
China's chip industry is very fast, from 2004 to 2018, plot development of the chip industry in China can be seen, we rose to 653.2 billion yuan last year from 54.5 billion yuan in 2004, $ 100 billion, this growth rate It is about four times the current rate of global growth. More than 6,500 yuan, in fact, is our design, chip manufacturing and packaging and testing industry three industry superimposed results.
We see the chip design industry last year reached more than 2,500 billion, which is the true sense of the product, and our packaging and testing industry 219 billion yuan and chip manufacturing more than 1800 billion yuan, this is more of a process. Then, design, packaging and testing, what is the relationship between these three chip manufacturing?
As an example, the design industry is the equivalent of writing a book writer, is equivalent to printing manufacturing, packaging and testing industry is equivalent to binding, their characteristics are not the same.
Our national enterprises, after after so many years of development, whether it is design or manufacture of IC packaging and testing have entered the world, such as in the global IC design industry among the top ten there are two companies, manufacturing companies in the world among the top ten there are two companies, and the global IC packaging and testing companies among the top ten among the three companies.
But we are compared with the international advanced level there is a considerable gap. Check out our design industry, that is, we often speak of integrated circuit products, from our industry-wide in 1999 only 300 million yuan, last year we have reached 251.9 billion, about $ 37 billion combined, have done the world's second largest .
While great, but if we take a look at our products in the world accounting for only 7.9%, if you still remember just when the film goes on, the Chinese market more than 1,500 billion US dollars, accounting for 34% of the global market, and this which we have only 7.9%, we have 26% rely on imports.
Some comrades are worried that we bought so many chips, if one day they do not sell it to me how to do it? It is not controlled by others, and so on, right? There is this fear is natural, but if we think about, as a manufacturer and supplier for what they will worry about it? They will be very worried. There was a foreign friend asked me, he said you bought so many of our chips, or else one day you bought, then how do we do? It will carefully smile.
This is a very interesting phenomenon, we do not sell our fear of others, fear of people we do not buy. So in this case, the best way is our own development, our own production more than a little, we all live in peace, this is best.
Fifth, the development of China's chip industry, which is facing problems?
(1) domestic chip industry and the gap between demand
In fact, the challenges facing the chip industry is very large, it is a huge project. We see from the perspective of the product, it should be said that we are the product mix between our needs and, there were some mismatch phenomenon.
There we see a lot of 0%, 0% does not mean that the absolute value of 0, is market share. Speak percentage market share of 0.5% or less can basically rounding, because you really attract any attention to people in the market, you must go to emphasize that I am not 0, it did not mean anything.
For example, say we probably annual Chinese imports of the CPU, there may be about 1 billion unavoidable. If even one billion of it, then we have a business that I produced 1 million, that's a lot, but an incredible one million things, but you put one million one hundred million with 10 to race with, you in fact, you know that is one-thousandth of only 0.1%, so we say 0.1% of the time you can not see in the market. So when we look at these things, not simply to see an absolute value, we look at its relative value, that is, market share, this is a very important point.
We can see that both the server or PC, or programmable logic devices, digital signal processing devices, as well as some of our core IP terminal, whether you use them, and some memory, what we are a large number of 0, This means that our industrial structure, we are designing the structure of the company's products with our needs there is a considerable gap.
The only two points that we see more than 10% or even 15%, that is, mobile communication terminals, this is the world we now relatively strong, accounting for about one-fifth of the global market.
(2) the development of lagging investment is not enough
Mismatch between design and manufacturing capabilities and development needs of China's chip industry. Our manufacturing spend a lot of money, but developed very rapidly, but still slow.
The most advanced integrated circuit manufacturers of our continent, when they are at 14 nanometers, about a quarter of this year (2019) can be put into production, while Taiwan's TSMC, their 16 nm, early in the fourth quarter of 2015 on the production. This is the middle there is a gap of three years, this is our place relatively than the others lag.
In addition to our fast enough, there is a terrible, our production capacity is not enough. If you can find the capacity, of course, you can make money, but you may not find the capacity, in time to grab the world's production capacity, production capacity you can not find how to do it? This time it is in trouble, then we will lose money.
We say that the development of integrated circuit chips need to invest, how much money they will vote for it? Astronomical! Global statistics on semiconductor investment, we see that except for a handful of years, most of the time in more than $ 40 billion in recent years even more than $ 60 billion. The red line is a piece of our national investments in semiconductors, in which the bottom.
Some people say that our investment is within the statistical margin of error of the people, listening to the words difficult to hear, we are also very hard to accept, why not invest in our country in it too? Our understanding of this industry is still limited, we made a relatively early decision or the wrong decision, that the Chinese semiconductor chip industry can be healthy development through market allocation of resources.
On this chart, the red line upturned recent years, a lot like Alice process. But we know that it is high strength industrial investment, whether Intel or just the Ye Hao Samsung, TSMC or, probably annual investment in billions of dollars of scale. We also went to the scale of billions, but we put a lot of home, your investment intensity is not enough, and just two or three years later to hit consecutive years to see results.
Now we say that the development of integrated circuits has become a thing of all Chinese people probably agree, it brought a side effect, is creating a big national integrated circuit.
IC is not a thing can be everywhere. I've come to a place where the local leaders say, we determined, and should make up integrated circuits, integrated circuit built here in our factory. I said, I'm afraid not, you have no money here. I said no money then, people very unhappy, immediately said how do you know I have no money? With the person next to say, I'll give you 50 million you gave me this thing done! I told him, I am afraid the back also coupled with zeros.
I'll give you an example, the US nuclear-powered aircraft carrier strike group, Nimitz-class, not now Ford-class, including a 100,000-ton nuclear-powered aircraft carrier, about 60-70 aircraft carrier, two guided missile cruiser, two guided missile destroyers, a nuclear submarine, as well as the supply ship all add $ 15 billion, we are now building a $ 15 billion integrated circuit plant. Therefore, the construction of integrated circuit factory, often require huge investment, and it was not a one-time investment, a lot of this pressure, this is not a little money.
We also have some local governments also very interesting, very enthusiastic on the integrated circuit, I can understand them for local economic development is devoted all their efforts. They see that, once established a factory integrated circuit, chip plant in their place, it will soon bring jobs, bring the surrounding environment, the ecology of the surrounding infrastructure, can bring a big industry, but they are difficult for chip development Getting To Know is small.
Once there is a place, not a local university Microelectronics, they have the chip industry, I said that you people from? He said that my people can "dig" from the outside, you let people sit four hours flight from Shanghai to fly you here, not realistic, right? First no one, no money second, third is not technology, so the development of integrated circuits that need a lot of investment.
(3) To strive for the chip industry chain
Development of China's chip industry also faces resource mismatch.
At present, our chip manufacturing over 50% of customers are overseas customers, our packaging and testing are also about nearly half of the customers are overseas customers, we are working for others. That our design industry is most in need of resources, but also all over the world to find resources to find resources processing, because the technical level of our manufacturing and packaging and testing industry, with our needs as well as distance.
Our original external processing industry is based, we know that "three to a meeting" and so on. This is the processing of industrial structure, now to become independent innovation, you have to make the adjustment of industrial structure. Central proposed to supply-side structural reforms, in fact, the chip is, we are faced with such a reform.
We in the development process, in fact, also facing the problem of an industry model.
Chip development has been for decades, in fact, in the development process, now as before is very different. In the past called the system vendor model, is to do everything himself, then said no, integrated circuits every 18 per month to double myself not working, called the emergence of the so-called integrated device manufacturing model, go down after the emergence the so-called design called a foundry model.
This is actually three business models brought about a different outcome. We look at the situation from mainland China, mainly design foundry patterns, foundry model that this design is good or bad? We go bad evaluation, because it is a historical stage of the decision.
In fact encountered in the real work of many of our problems is that many local governments are very keen to build a factory integrated circuits, integrated circuit as a factory to spend billions of dollars every turn tens of billions of yuan, the local GDP contribution is significant.
But it tends not to think that a country like China, always working to do things that middle and lower reaches of the industrial chain, then you are not put himself in the frame of the middle and lower reaches of a chain? Therefore, the central government proposed to us to innovation and development. Where innovation and development? In the upstream, we so often say we want to walk up.
(4) lack of talent
We want to develop its chip design industry, chip design is a high-tech talent has become an important constraint.
Where bottlenecks encountered in it? Not only quality is difficult to meet the demand, and now even the number is difficult to meet the demand, the most direct effect is that we are now at each other the entire semiconductor industry to dig people. If you are engaged in the chip, now quit will be able to find a good income, because you can pay double.
We made a statistics, mainland China engaged in chip design engineers, has been higher than average pay in Taiwan. Tell the truth, we might as Taiwan's ability to engineer, what does this mean? Our products are not made out of good people, but our costs are higher than the others, this situation has not eased, so we have a shortage of talent team is very scary.
Two years ago, we had a sort of a small trouble in personnel training, is that many students engage in investment after graduation, engaged in finance, of course, my own students have to go out to make investments, do officials.
I always say, if this is the case, why do you want to come to learn so many years of integrated circuits it? They also say the importance of the chip, for this infinite charm chip itself contains enough to understand, he just put it as a knowledge to learn.
When you really understand chip, integrated circuit after its internal things, as well as its external influence such development, then you will know, the original master integrated circuit chip can bring so much initiative.
Demand, supply shortage is a challenge we are facing China's chip industry, this is a real problem, is our next step as the point when supply-side structural reforms of a concern.
If we engage in future chip technology, I believe from my speech today among the least grasp several important points:
The first point, I think the development of the chip is probably not the people's will, keep going, but also grow a hundred years;
The second point, the development of the chip is not easy, is not so simple, requires high investment, we need long-term adherence. A hundred years is not just a number, a year, but that will have long-term adherence results.