Intel has been shining for years in Silicon Valley, where technology companies are shining with stars. On July 24, the second quarter results showed that Intel had revenue of 19.7 billion US dollars and net profit of 5.1 billion US dollars.
However, the beautiful report of the second quarter didn’t bring about a sharp rise in the stock price. On the 24th, Intel’s 7Nm process chip jumped again because of yield problem. In addition, venkata, the key figure of Intel, also left the company. The stock price of Intel once fell by 10%, and the market value evaporated by $25 billion.
in sharp contrast, AMD, the rival that released financial reports, had a revenue of less than one tenth of Intel. However, intraday trading rose by 10%, and its market value even surpassed Intel for the first time.
from Moore’s law, which created the golden rule in the field of chips, to the “Pentium core”, which Bill Gates called “the king of chips”, Intel has been riding the dust for decades. TSMC and AMD were once surrounded by Intel. Now, the two processes have already moved towards higher processes of 7Nm and above.
it’s not embarrassing for Intel to skip the ticket for chip manufacturing technology this time, which has happened before. More importantly, Intel announced to the outside world that, in order to be “more pragmatic”, it acknowledged the rumors that it would consider outsourcing part of its chip production to a third party.
Intel and TSMC have always been in a cooperative and competitive relationship. When TSMC was just established, orders from a part of Intel’s backward manufacturing process gradually made TSMC out of the loss state and subsequently was recognized by large manufacturers.
However, the cooperation between the two sides this time is not quite the same. According to media reports in Taiwan, Intel will begin mass production of 180000 processors or graphics chips using TSMC’s 6nm process next year. The technology used is also one of the most advanced technologies of TSMC.
TSMC and Intel have developed from low-end OEM to high-end cooperation today. In this fierce battle of chip manufacturing process, the latter has been surpassed in literal process number.
the smaller the circuit width carried on the chip, the faster the processor runs. However, due to the limitation of chip area, if more circuits are inserted into the square, the diameter of the circuit will be smaller and the process will be more complex.
the process nodes of
chip have gradually developed from micron (μ m) level to 32 / 22 / 16 / 10 / 7 / 5 / 3nm (nanometer), and the common half node is 28 / 20 / 14nm.
as for the evolution law of chip circuits, Moore, one of the founders of Fairchild Semiconductor and later the founder of Intel, put forward the “Moore’s law” which is still used today as early as 1965: “the transistors integrated on semiconductor chips will double every 18 months.”.
Intel’s first processor, the i4004 processor, is basically a practical test of Moore’s law. During this period, Texas Instruments and IBM also launched related products.
it was not until 1993 that the appearance of Intel’s iconic “Pentium chip” helped Intel become the world’s largest semiconductor company in the following decades. This process war really began – it was also the year of TSMC’s turnaround.
in the late 1980s, under the background of the trade war between the United States and Japan, the semiconductor industry in the United States began to shift. Under the attack of the United States on Japan, the semiconductor industry in Taiwan got a start-up opportunity.
TSMC started under this background. However, in the first few years, TSMC’s production capacity was insufficient, the yield was too low, and TSMC had been in a loss state.
now known as the “king of chips”, TSMC founder Zhang Zhongmou took part of the list from Intel at that time. At that time, TSMC, which was several generations behind Intel, was not enough. Under the direction of the United States attacking Japan to support Taiwan’s industries, Intel also helped TSMC produce 0.8 μ M technology. With the endorsement of Intel, TSMC also won the orders from Italy France semiconductor, Broadcom and NVIDIA in the same year, becoming a real chip foundry.
Intel is not idle. “Pentium” CPU and Microsoft Windows 95 realize the binding of software and hardware, and the win Tel alliance is the most powerful combination in the era of personal PC. After that, Intel launched a new processor brand named “Xeon” and “Celeron”, which entered the field of professional processor / low-end processor respectively, which eroded the PC market in an all-round way. With Microsoft’s support, Intel’s chips have swept the world and become the largest chip manufacturer.
there are three production modes in the semiconductor industry. One is the fabless mode represented by Hisilicon / MediaTek / Qualcomm, that is, the chip suppliers without factories are only responsible for the design and sales of chips, and outsource the production / testing / packaging links required for chips; the second is the foundry factory mode represented by TSMC, which echoes with the former and is undertaken by the companies of this mode Manufacturing tasks of several design companies.
the third is IDM mode represented by Texas Instruments, Intel and Samsung, which undertakes all aspects of chip design, manufacturing and packaging. In the early development of integrated circuit industry, enterprises generally use this kind of IDM mode.
However, core making is too difficult and expensive. IDM mode can grasp the overall situation, but the return rate of construction production line is too low. In recent years, in addition to Intel’s insistence on self-study, self-production and self-sale of chips, other companies can’t play without sufficient funds. Semiconductor industry is bound to move towards fine social division of labor.
both TSMC and Samsung are strongly dependent on the technology of American semiconductor companies led by Intel / IBM. Similar to the Varna agreement, IDM manufacturers can not disclose the latest generation of technology to the OEM manufacturers in order to ensure their leading technology. So TSMC could only contact part of Intel’s OEM tasks with relatively backward manufacturing process.
in the game, the OEM companies want to be the accessory of IDM manufacturers, not the vassals. In order to achieve faster technology catch-up, TSMC is also in-depth cooperation with other major IDM manufacturers to get rid of its dependence on Intel technology.
in 1998, TSMC put forward the “mountain plan” to meet the actual order demand of five semiconductor giants, such as Texas Instruments and Motorola, to adjust its own technology upgrading and enhance its production capacity. This has also helped TSMC form a positive cycle – better understanding of the upstream demand can better produce orders. It also has the capital to continue to polish the follow-up technology and improve the yield rate of the production line, so as to obtain more orders and continue to share the high technical costs.
TSMC has become the backup force for the major IDM manufacturers in production. With many efforts, TSMC and Intel also aligned in the 0.18 μ m process for the first time in 1999, which is still on paper. At this time, TSMC was still in the catch-up stage because of yield problems.
later, TSMC’s rise was inseparable from the decision of Jiang Shangyi, the person in charge at that time, and the support of ASML, a lithography equipment manufacturer born from Philips.
45nm in 2002 and 5nm today are two major barriers in the development of semiconductors. The breakthrough needs to use ASML immersion lithography machine and EUV respectively.
ASML’s first immersion lithography machine was jointly developed with TSMC, and the research and development of extreme violet lithography was also jointly funded by Samsung / TSMC / Intel. The former helped all manufacturers including Intel to overcome the 45nm mass production test. Under the background that most manufacturers choose to abandon the R & D of advanced processes above 7Nm, TSMC also has the most EUV resources ASML is also the first manufacturer to help TSMC make rapid breakthroughs in advanced manufacturing processes.
at present, the main CPU is the vertical division of labor mode of fabless + foundry. The competition between Intel, NVIDIA and AMD is actually the competition between Intel as an IDM and TSMC. In recent years, Intel has adopted third-party OEM more and more times, which is not only because of capacity problems, but also because of technical factors, which also shows that Intel’s IDM model is tearing up and restructuring.
before that, other IDM manufacturers have successively reformed. For example, amd spun off the CPU production business in 2009 and established Grosvenor (grid core). IBM has gradually sold a number of “less important” businesses in 2014, including the wafer plant.
an interesting example is that at the 30th anniversary celebration of TSMC in 2017, senior executives of apple, Qualcomm, Broadcom, NVIDIA, anmou, ASML and other companies all came to Taiwan and picked up one of them. They are now heavy players in the industry chain.
in 2007, Intel summed up the “tick tock” mode. This name comes from the sound of the clock and the second hand. Each tick represents a process update, and a tick represents an architecture improvement.
a tick stock is a cycle, which takes two years, and basically corresponds to the speed of the next generation process in 18 months of Moore’s law.
2014 is the year when Intel and TSMC really opened the gap. At that time, Intel entered the 14nm era, but it decided to “squeeze toothpaste”. It adjusted the two-year cycle to three years, and inserted a one-year optimization period on the basis of the original “process technology architecture”.
but in practice, the advanced cycle of Intel’s chip manufacturing process has been lengthened again and again. The 10 nm process originally planned to be launched in 2016 was finally delayed until 2019, and the 7Nm process technology that should be launched in 2019 has been delayed again and again.
at the same time, the 10nm processes of Samsung and TSMC were all mass produced in 2017. The latter was first used on the chip of Apple a11 (with the iPhone 8 / 8plus / x), 7Nm in 2019, 5nm in 2020 and 3nm in preliminary progress.
Intel is also aware of this, so it has done a lot of work to improve transistor density. Some scholars also pointed out that the density of transistors in Intel’s 10nm process is twice that of TSMC and Samsung, and is about equal to TSMC’s 7Nm density.
while Intel is working for OEM, TSMC’s capacity, especially the capacity of 5nm process, has been occupied by apple and Huawei Hisilicon in advance. Intel’s challenge is how much more 7Nm process technology can rely on its own factories and how much capacity assistance can be provided to Intel by a third party when the capacity of the previous generation chip process is in short supply.
Intel still has a lot of ways to go. Whether it’s storage, traditional PC business, or new business like data center, Intel is no doubt the overlord, and has left pieces in the Internet of things and mobile eye.
many people have also decided that even if Intel regains the established rhythm of Moore’s law invented by itself, it will not be able to catch up with TSMC at 5nm / 3nm. However, between Intel and TSMC, there is a competition between the former “king of chips” and the “king of OEM”, which is a speed competition of advanced manufacturing processes in the semiconductor industry and the status of the world’s advanced manufacturing industry.