Intel's billion gambling tomorrow technology is worrying, but the chip giant Intel is a factory, and it is a new market. Intel bet is: Let all competitors can not hold a candle Author: Brent. Brent Schlender, is the intelral D1D semiconductor plant completed in Hillsboro, Oregon, and its amazing complexity is comparable to the upcoming microprocessor structure. Entering the plant's wide underground layer (total value of $ 2.5 billion), just like the core swimming of computer chips: thousands of workers and technicians seem to be used in electronics between chip aluminum wires and copper wires. It is busy operation in this huge underground concrete workshop to test complex high voltage circuits, connect the lines and pipes of the meandering miles, and install external support devices for the above two-layer assembly area. The area of the front room is so big, the football field is so big, the light can be observed, dozens of design experts are wearing dust-proof clothes, wearing a hard hat, and a meticulous busy. Some of them are carrying precision chip making machines, they put these equipment with the genitals of the ice machine to the pulley, then slide on the compressed air cushion. In the spring of 2003, when D1D was put into production, it will become a "factory" in the world's wafer production. The dimensions of these wafers are comparable to the medium-sized pizza, each with hundreds of advanced microprocessors, each with nearly 250 million transistors in each microprocessor. But D1D is not just a synonym of Pentium, which will be the latest and realistic manifestation of Moore's Law. Moore Law is one of the founders of Intel. Gordon Moore made 37 years ago and has been leading the development of the entire semiconductor industry. He took the lead in pumet a time, the number of transistors and electronic components on the chip increased doubled. At the same time, D1D is also the Declaration of Intel's own, and its mission can be understood as "opening factories, production." To this end, Intel has been a specific embodiment of this goal for establishing a strong "Moore Wall", D1D, and other factories planned in 2004. Intel let go, it is to develop first-class chip technology at the speed of super competitors, with a powerful manufacturing advantage, in terms of price, performance and species. This means that Intel will continue to work hard in advanced micro-equipment, but also continue to move the top of the computer chip manufacturing business; at the same time, we must also go to Texas Instruments, IBM in all the other chips (from mobile phone chips to automotive chips). Motorola, as well as many other smaller competitors to challenge. Intel's former CEO, current chairman Andy. Andy Grove said: "The scale is a strategy. Henry Ford is a revolution in the automotive industry with this strategy; Japan has defeated us in the memory chip in 25 years; It is also a revolutionary revolution with this strategy. Now we have once again enabled this strategy to extend the business from the computer field to a broader world. "Intel is struggling to get rid of the most severe dilemma in history in IT. Affected by the atrophy of computer industry, the company expects the 2002 income only $ 26 billion, 25% lower than two years ago. The company's sales prospects and the largest customer size have no sign of rapid improvement. At the same time, another consumer market for semiconductors - the telecommunications industry is almost stagnant.
Although Intel is still profitable, this year's net income is expected to reach $ 3.3 billion, but earnings have been reduced by 8% over last year, which is reduced by 71% compared with 2000. Intel stock prices also dropped 80% over $ 74.87 in 2000. Intel's competitors were more difficult, and many of the income deficits were there, and the plan to build a new plant was also put on hold. AMD is currently facing severe cash pressure due to unexpected losses. Other chip manufacturers (such as Motorola, etc.) announced that they will completely give up the intention of manufacturing chips, they will hire the so-called chip aggress of Taiwan to make chips. This means they need to wait for a longer time to improve performance and can only use non-first-class processing technology. This is also a reason for Intel vigorous expanding production scale. Intel believes that by conducting a large number of technological investments during the depression period, the company can lead the leading advantage in chip technology and manufacturing capabilities. CRIG BARRETT: "Currently, the integration and differentiation of the semiconductor industry has just begun, it will eventually evolve into a small number of leaders leading a lot of followers to advance. And in terms of product and profits There will be corresponding differentiation. Only those products with first-class design and performance can get the highest profit, and only this point can be achieved by using the cutting-edge manufacturing technology. "Mr. Bayer is entering the open office of Intel headquarters in California Clara. Mainly responsible for chip manufacturing. This is a typical strategic declaration for a company-known company, perhaps like a typical strategic declaration. But this bet that Intel's Moore's law can bring a second opportunity to it. It actually represents Intel's unwavering belief in technology development, as well as the desire to bring dawn for IT and telecommunications industries. Intel firmly believes in the ability to make more powerful chips, enhance the Internet to a new height, so that many computers, phones and other devices can connect to Wi-Fi. It is likely to succeed in paul otellini, which is a President and Chief Operating Officer of the next CEO, "We now hope to integrate 500 million" or even 1 billion transistors on the chip, soon you will be designed No restrictions from a single integrated circuit. At that time we will enter the golden age of the semiconductor industry, and IT, telecommunications, and home appliance industries will become the biggest beneficiar. "But there is no sign to ensure this era. Assembly, if it fails, Intel's profits and share prices will fall sharply. Even if Intel can continue to expand the market share of 81% of computer microprocessor market, it is still not able to bring sufficient sales to use this superior production capacity, and it is also impossible to reproduce the growth records of two digits. Although the telecommunications industry may benefit from Intel's innovation, even after it recovers the vitality, the number of chips required can push Intel's development or unknown. The industry is full of suspicion, especially in Wall Street. Jonathan Joseph in Salomon Smith Barney said: "Intel has a fear of production capacity. They are very worried that they will miss the right opportunity to reverse the situation. But obviously their production capacity is now excessive, they I hope to be able to hit the target, but I find that the goal is not there. Intel will need to adjust to maintain a profit, which means they will have to close some factories. "Moore's law also has certain cruelty: it predicts, no matter the company, the whole There will be any situation in the technical industry and the economic environment, and the technology will continue to grow exponentially. For chip manufacturers, slowing innovation will be exhausted. Mr. Moore has shown this view in "Electronics" magazine as early as 1965, and he is still confident that he is 73 years old. He said: "I am willing to guarantee any problems of Moore's law.
"The law was originally only a simple speculation of Mr. Moore, mainly to explore how fast the emerging chip industry can accommodate more components in a single integrated circuit. [Carver MEAD, famous physics, California University Later, it was called Moore Law. He thought that "it is a personal prophecy, not only the law". In 19975, as the chip became more and more complicated, he adjusted the growth cycle to two years. Recently, Intel increased double the number of transistors to double the cycle to 21 months. In limited space, the number of transistors increased by one Double, only the genius design is not enough. Chip manufacturers need to cooperate with semiconductor equipment manufacturers to develop tip manufacturing techniques, narrow the volume of transistors and other tiny functional components to one of the original, in chip To integrate more components. This requires more advanced lithography technology to accurately integrate or move micro-metal metal and other substances on chip wafers, and the chip to establish the most net factory. D1D The diameter will be less than 90 nanometers (only one tenth of people in diameter), far ahead of the current 130 nano minimum chip diameter. Each generation of chips need to establish a new plant or completely refurbish the original factory Therefore, with the increase of Intel's investment, the cost of the annual factories is also steadily climbed like Moore's law. But its return will also be very good. Higher chip density means higher performance and capacity; a first-class new chip The factory can reduce the cost of chip to the original one. In addition, smaller volumes can often make performance higher. Mr. Mo said: "This is a very interesting technology. After all things are smaller, their performance will increase. But this also means that if your technology is backward competitors, then it is not as simple as the chip performance, you have to cut costs. "The following sentence really reflects the ruthlessness of Moore's law. Intel Chief Financial Officer Andy Brient said:" Every time we don't realize the law of Moore, others will. Once you fall, you will suffer more disasters. "Berett added:" This is what we must insist on driving the development of Moore's law. This has become the basic principles of all Intel believers. The first class of new employees we have moved to the company is Moore's law, telling them: 'Hey, our predecessors are in the second, and ten years ago, I speculated this law, I think you can do it. 'We resolutely oppose difficulties, and increased complexity as an excuse that does not push this Moore's law. "For Berrett, Moore's law has become an important criterion that he insisted on the use. People responsible for the actual implementation of Moore's law is SUNLIN CHOU, a 56-year-old Confucian, and he worked in Intel for many years, now Advanced Vice President and General Manager of Technical and Manufacturing Division. Berett allocated a large number of resources to work, including nearly half of the Intel R & D budget (this expenditure will be as high as $ 4 billion), used to develop will be in D1D The new manufacturing method is enabled in the factory. (The amount of R & D funds is the number of Intel's other years to maintain near-index growth.) Zhou Shanglin's efforts are the never-ending. D1D's 90-nanometer factory even has not completed production, he squatted Researchers have begun to design new manufacturing methods, and work closely with equipment suppliers to achieve the next generation of Moore's law. By then, the chip characteristics will be reduced to 60 nanometers, while Intel hopes to start in the factory in 2004. Use this process. In addition, starting from 90 nano-factory, we will introduce multiple processes to enhance the plant to get rid of a single mode of a particular type of chip.
The processor, flash memory, and communication chips are using a slight difference, but the new plant will be able to produce all of these chips without lowering total production. Zhou Shanglin will also use 90 nm to make a larger wafer in the Intel's new factory. The wafer size is generally expanded in ten years, and many chip manufacturers around the world will suspend this measure to save costs. From now on, Intel's new plant will produce a diameter of 300 mm (about 12 feet), while the existing 10 factory has a diameter of 200 mm (about 8 inches). Zhou Shanglin explained: "The size of the chip characteristic makes the process more complicated, causing the cost to increase, while the expanded wafer size can offset these costs. It is entirely due to economic considerations. In addition, it can also produce it on a wafer The number is the previous two-fold chip, thereby further reduces the cost of a single chip. "One of the most important responsibilities of Zhou Shanglin is to ensure that the D1D process and factory design are indiscriminately applied to other 90 nanometers, the latter. It will be put into production in the next 18 months (a in New Mexico, one in Ireland, another in Oregon, belonging to the old factory transformation). This method called "precise replication" technology is the first in Berrette, who is the first large factory expansion process, and its purpose is to ensure that each new plant can be put into production immediately after completion. The secret of precise replication strategy is that it is common to restrict the design engineers' creativity. Zhou Shanglin said: "Once we have determined a set of processes, we will not let the chip design team will change freely. I am telling that strict discipline regulations can guarantee that the designer does not change, because this does not meet the nature of the engineer. "Zhou Shanglin's hard work has achieved this final result: With the successful investment of D1D and its similar plants, Intel's production capacity will double. Intel can not only increase the chip production capacity, but also the chip that has a significant improvement of performance - Pentium will increase from the current 2.4GHz closing to nearly 5GHz - and the average production cost is reduced below. Other important products of Intel will happen, such as cellular phones, handheld devices, and digital cameras, and the company hopes to drive new rounds of growth networks and communication chips. In other words, I will see if Intel can take advantage of the competitive design, and if you can find your customers who are interested and have the ability to buy. The Intel Communications Division believes that even if the telecom industry is still unable to return to the day, Intel can still be a big offense and in the development of the new market process, Intel may find that the Moore wall is like a pyramid as a pyramid. If no one is purchased, the technical advantages of the new factory will talk about troops on paper. Wall Street will not return for good wishes. This is not to say that Intel has no opportunity to expand the current 80% share of 80% of its income. Due to the increasing popularity of Linux and Windows Software Incoming more and more popular in business calculations, Intel Pentium 4 and to Strong Microprocessors have been continuously expanding market share. Intel new high-performance processors for high-end servers - Anteng processor so far, so far, it is expected to have a large development after its 90-nanometer-model product is available (the unit price of up to $ 4,300 is able to bring Huge income). Then there is mobile calculation. The sales of laptops continued to grow at 15% per year, and Intel is preparing to place more chips into every laptop. Intel is highly hoped for communication chips and will deepen it as the driving force for promoting enterprise development. These chips allow laptops to wireless communication with the network, allowing cellular phones to call, send web pages, email, and streaming media via the Internet. Intel firmly believes that the calculation and communication is integrated (actually achieved through the chip itself), it will be a business success.
Chief Technology Officer Pit. The Pat Gelsinger named this strategy as the Radio Free Intel policy. Briefly, he hopes Intel to automatically detect and connect to the new Wi-Fi wireless network or even the radio transceiver that is completely integrated into its various processors. Kissinger asked: "How can we defeat the experience in communication technology than us in Texas Instruments or Motorola, which makes companies such as Texas Instruments or Motorola? Only through changing rules, a new architecture that integrates communication capabilities in intelligent devices. We want to create a radio transceiver that has any equipment with a microprocessor. "Intel wireless enhancement processor is the most likely marketing computer. At the end of this year, Intel will begin shipping some chipset samples designed specifically for laptops, including ultra-low-voltage pentium processors, graphics chips, and other support circuits, and built-in chips connected to Wi-Fi networks. These chipsets will allow the laptop to detect and access the wireless network when they are out of the owner, even switch from a network to another when the user rides a plane. Chandrasekher, Vice President of the Supervisor Product Line, said: "In the field of mobile computing, if we are still in the process of the processor, I am afraid it will lose the target. 窍 is hidden wireless communication All additional properties needed to make it play behind the scenes, allowing users to be convenient and reliable. "The second largest target of Intel wireless freedom points to cellular phones and PDA - this is Intel has been involved but has not yet occupied Market market. This year, the sales of cellular phones will reach 400 million, including many of the Intel flash chips. But the phone is also becoming more intelligent, but also starting similar to the PDA on the processing address book, calendar, etc. At the same time, Intel's XScale processors are most PDA brains (these PDAs use Microsoft's Pocket PC software), recently won the Palm support. It may become an industrial standard, just like a Pentium a personal computer standard processor. Intel's grand plan is to combine the Xscale chip with flash, introduce more chips into a cellular phone in this way. It also plans to use the same part included in the new Wi-Fi chip to make PDAs a more comprehensive communication device. Intel hopes to eventually support communication transceivers for Wi-Fi and voice cellular telephone services, XScale processors, and large amounts of flash memory - integrate to a single part, as a cardiac component of PDA or cellular phone. To achieve this goal, depending on whether Intel can create a chip containing a smaller transistor, can you master the integrated wireless components, logic circuits and memory in the same chip package, while maintaining electrical signals that do not interfere with each other. . In the process of entering the cellular phone, Intel will have to connect with many powerful competitors in this field. For example, the leader of the global communication chip, Texas Instrument, monopolizing more than half of the cellular phone processor market. A spokesperson of Texas Instruments said: "When other companies are Intel, you can't ran away immediately when you are interested in entering your business. We have a long-term good cooperation with cellular phones - handheld equipment manufacturers. "Moreover, the success of the microprocessor business mainly depends on the continuous optimization of a complex chip and a little support circuit, and the larger manufacturers of communication chip services are mainly obtained by developing and supporting many different types of special chips, many chips. It doesn't really need the latest and most advanced process. Therefore, the problem is that these chips are very profitable, far from the profits brought by Pentium. Intel enters the communication chip market and another main purpose: occupying the market of infrastructure that telecom operators and equipment suppliers are widely deployed.
In the past four years, Intel has sold $ 10 billion to acquire 27 companies related to network technology, from network processors to wireless technology and optical exchange systems for fiber optic networks, etc. Now Intel is working to produce a network processor, which will become "Pentium" in the telecommunications field. The network processor is a super fast switching circuit located at the core of the Internet digital data routing device. So far, there are no network processor standards, Lucent, Nortel and Cisco (Cisco) and other companies are purchasing specially designed chips and build their own software tools. Every company wants Can win more in exchange performance or programmability. Another reason for this proprietary manner is that the customer tends to supply long-term supply after familiar with a supplier's programming tool. Sean Maloney, which Intel Communications Division, Executive Vice President, believes that this chaotic situation must end. "I am not saying that competition is not good, but the telecom industry has never really benefited from Moore's law. The reason is that most of these proprietary network processors have insufficient production, and cannot guarantee the application of cutting-edge manufacturing technology. Standard parts produced by first-class manufacturing technology Plus a standard programming tool evolved from the personal computer software tool, the telecommunications industry can ultimately introduce Moore's law. "The Intel Communications Division recently announced a high-performance network processor chip -IXP. It is designed as key components in the router and digital switches, which can process up to 6.6 gigab.com, email, streaming media, query, and other Internet traffic per second, which is approximately equal to 4,500 household broadband lines. But the future is not a bright. The current telecommunications industry is overall, and the market transaction is bleak. As of September 28, the Intel Communications Division of Intel Communications Division was only $ 482 million, only 7.4% of Intel's total sales revenue. Worse is that this sales number fell by 17% higher than the first quarter, a decrease of 10% from the previous quarter. Such sales performance is difficult for Intel Expanding business or growth. Bryter, CEO, admitted: "From the current situation, our previous expectations are too high. Although the performance begins to grow more is related to the macroeconomic situation, not technical issues, but we I believe this day will always come. "Optimistic Ma Hongli (he once in charge of Intel Sales and Marketing Division) Thinking Intel can see a warmth in the" cold winter "in the telecommunications industry. He said: "In the past two years, our telecommunications customers have cutly cut 270,000 employees, many of which are engineers designed with special network equipment, and telecom companies will never have a strong engineering and technical force. It will be responsible for developing core components by us, and they will focus on special marketing equipment and services so they can benefit from Moore's law as the computer industry. "Gruf said:" I heard that the saying 'Is the Internet change everything? I want to say, wait for five years, "This optimistic manifestation is in Intel may make people feel strange, because Intel has always had an incumbent point of view:" Only paranoiders Can survive. " But confidence in Moore's law is rooted in the deepest place of Intel's spirit. In the so-called authority, it is repeated that the chip process will reach the vertex as the circuit size is reduced to atomic magnitude, Zhou Shanglin and its peers in the chip industry are constantly finding new methods to successfully integrate more in one chip. Transistor. Zhou said: "Market opportunities are not shrinking, but in explosion. In the final analysis, Moore's law is not just a strategy, or a philosophy. It gave us a confidence in the future." Take a look at Andy. Gruf, this Hungary immigrant who has experienced countless brilliant and setbacks is how to see the dark days he has encountered. He said that you must have confidence, rationally judge the technology trend, just like Moore 37 years ago.
In its original article, Moore successfully foreseen that "Integrated Circuit will make household computers, automatic control car and personal portable communication equipment", and at this time, the phone is still a dial, color TV is a new thing. . The following is the forecast of Andy: "Have you heard of the 'Internet changed the world?" People now be a little shake, but I have to say, wait another five years. We now spend a lot of millions of dollars. The voice communication service will become a free service, just like Chambers (Cisco CEO). This is the utility of Moore's law. The entire entertainment industry will provide products and services in the broadband Internet. Media Company] The sky-covered changes will occur, because one company will have to follow it as long as one of these companies has achieved transformation, other companies will have to follow. This is the utility of Moore's law. Home will implement wireless connection, broadband The service will also implement wirelessly, home and portable computers and other household appliances will implement wireless connections. Even if it does not achieve five years later, it will become a reality ten years later. This is of course for some people. Not good news, but as this will be achieved, and we will benefit it. "This is why I am worried about Moore's law is too unrealistic when I am too unrealistic, and Intel is still convinced that this is the only way. . Translator: Chen Chunmei ...
How does Intel becomes the "Moore Law" as a belief?
In October 2002, "Wealth" reporter interviewed Gruf and Moore, the theme of discussion is Intel's early entrepreneurial experience. The following is a summary of this interview. In 1965, you proposed a discovery that we now call "Moore Law" now. After the year, when you with Robert. When Northis and Andy jointly established Intel, would you realize the "Moore Law" as the company's clear goal? Moore: You know, in the early stages of Intel's development, the company did not regard "Moore Law" concept as a driving force. In fact, until the last 10 to 15 years, we have begun called "Moore Law". The original intention of our establishment is just to expand the storage circuit technology towards the right direction we believe. Gruf: We didn't incorporate it into our company's mission, we just let them naturally. Moore: Our goal is to work hard to develop as high speed as much as possible so that it has market value. And facts have proven that technology development is roughly followed by a certain law. We really start driving its development is when developing a few generations of technology. Although this challenge has become more difficult, and it is necessary to pay more, but in fact, we have increased its speed, which is really surprised. When we establish and start the 90-nanometer factory, we believe that we can launch next to the next generation with a faster speed than "Moore Law". I have witnessed technology growth in the past few decades. I have it to believe that we can design or manufacture these products. Why Intel's "Accurate Replication" strategy - enables new plants to quickly achieve mass production - is an important factor in realizing Moore's law? Moore: This is critical, because in the life cycle of a generation of products, it takes a few years. From the past experience, if the speed of mass production is slow, it takes a long time to put a new technology into sufficient production. While shortening the renewal cycle of the product, we can also achieve mass production in advance. This way we can innovate faster, leading to competitors. Gruf: An ideal situation we look forward to this is that all old and competitive products and technologies can be transformed into new generation products and technologies through an innovation. It is unable to achieve changes to the new generation of products through functional upgrades, however, our fast "precise replication" mode can achieve this goal. Moore: The key is to transform the entire product line (not just some part), making it more competitive performance, perhaps just a few price points. It is often talked about how Intel has reduced prices by 20% or 30% or less - the rules of the game rules. When Intel will put new and higher performance chips to the market, the original top product is well, and their prices are different from our products. Our intention to launch a new generation of products is to improve competitiveness through new products with more performance, rather than reducing prices to improve competitiveness. Moore's law has brought you a return, and each generation of chip is designed to accommodate more transistors so that you can integrate more features into a base processor. But is you constantly increasing the number of transistors? Moore: Moore: In this process, we have tried a lot of programs, some of which have some halfway. Other components on the computer motherboard have also changed significantly. Thus, if you integrate all the features on the main chip, it will eventually lead to a significant decline in the flexibility of your computer, and flexibility is critical in the development of the computer industry. A few years ago, we have developed a product called TIMNA. The development of this product has made great progress, and graphics can be tightly integrated into the processor. However, when developing approaching end, we confirm that the market is not optimistic after the product is put on, because it does not have flexibility. So we gave up the product.
