Global Foundries, a well-known American semiconductor company Global Foundries, announced that it will abandon the research and development of 7-nanometer technology in the future. This means that Intel, Samsung and TSMC will be the only three major chip manufacturers participating in the 7-nanometer war, but Chinese people do not have to Frustrated, SMIC once stated that it will enter the development of second-generation manufacturing technology next year, and may go straight to 7-nanometer technology and grab the fourth ticket in the world!
In the eyes of outsiders, semiconductors seem to be a "stable profit without losing" industry, but behind the scene is crazy money burning and technology research and development. Samsung and TSMC must burn down in order to maintain their first-tier status, but second-tier companies such as UMC and GlobalFoundries can already guarantee that they can make money if they stop at the 14-nanometer process generation. If they rush into 7-nanometer, they may not even get it. At that time, it will be dragged to death by big companies. Intel has also faded out of the foundry business. In the future, even if it develops 7-nanometer technology, it will focus on its own products. Only TSMC and Samsung will be able to open up foundry services.
Why are companies desperate for research and development? Why can't domestic semiconductor companies develop by leaps and bounds like the Internet?
In the past, the winning strategy for semiconductor winners was to keep investing all the way without looking back and gritted their teeth. As long as they last longer than their competitors, the winner takes all in the end, because the chip industry is the first to make big money, and the second to beat Ping, the third place lived very hard, and after the fourth place all lost money, and because of this, all manufacturers gritted their teeth and continued to invest in order to squeeze into the top three.
It's different now.
In terms of 7 nanometers, research and development requires a huge amount of money, but there are very few users who can use 7 nanometer technology. Small and medium-sized companies do not use 7 nanometer high-end technology, and 16 or even 28 nanometers is enough. So you have never heard of Texas Instruments moving on 7nm.
Other people's microcontrollers and automotive electronics can do very well, and they don't need the most advanced manufacturing process. Even 55nm is ok! To give the simplest example: In addition to the processor is the most advanced product on the mobile phone, the other chips are all made N years ago, and they can be used very well. Can cars and home appliances use 16 nanometers? Therefore, for profit, other companies don't seize the technological highland, and through their own research and development and channels, they can also live very well.
Why the smaller the process, the more difficult it is?
The principle of the lithography machine for making integrated circuits is roughly as follows:
1. Apply photoresist on the wafer and cover it with a mask.
2. When irradiated with visible light, the irradiated part will react, and after development, the irradiated photoresist part will dissolve
3. Put it into hydrofluoric acid to etch, the part without photoresist will be etched, layer by layer.
Image source: Chaoneng.com
The reason seems very simple. But "light" is not simple.
The limit of the process is that light needs to pass through the mask. The smaller the process, the smaller the mask. To a certain extent, the problem of light diffraction must be considered. Among visible light, violet light with the shortest wavelength (wavelength around 430 nanometers) is ultraviolet light, which is shorter than violet light. The previous generation lithography machine used deep ultraviolet light (190 nanometers), and the new generation of extreme violet light (7 nanometer wavelength!) So if we want to break through the 7-nanometer limit, what kind of light should we use? No one knows. This is why it is often said that "Moore's Law will fail".
Extreme ultraviolet lithography machine (also known as EUV lithography machine)
Now you know how Tsinghua Unisplendour's name came from?
Back to the topic.
Only Apple, Qualcomm, Broadcom, Nvidia, AMD, MediaTek, HiSilicon, these "monsters" that still need a lot of performance to fill are the customers of 7nm technology, but these needs can fill the investment cost of 7nm? (Remind the latest technology, the monster-level RTX2080 graphics card just released by NVIDIA uses 12-nanometer technology, the previous generation 1080 is 16-nanometer technology)
Only TSMC can enter mass production of 7-nanometer process this year, and Samsung will join next year. Intel's 10-nanometer technology has been difficult to produce, and it has been postponed again and again. Even technology giants such as Intel are showing their difficulties, let alone others?
SMIC
However, the rise of SMIC will not make TSMC Samsung too comfortable in the future. It has just announced that SMIC, which has successfully developed a 14-nanometer process, is about to develop the next generation of technology.
People in the industry believe that after the successful development of 14-nanometer technology, SMIC is planning for 10 nanometers? 7 nanometers? Or is it an intermediate process technology? Although the company has not disclosed it yet, it is believed that SMIC will not be absent in the 7-nanometer process in the future.
Pure Sine Wave Inverter,Off-Grid Pure Sine Wave Solar Inverter,Power Inverter For Solar System,Pure Sine Wave Pv Inverter
Zhejiang Kaimin Electric Co., Ltd. , https://www.ckmineinverter.com