"No "independently created core" is not a technology giant. After the domestic self-developed chip is still in the stage of "localization rate" and "ZTE incident," the chip has once again become the focus. This incident not only alarmed ZTE but also had a profound impact on other companies.
Alibaba announced that it has researched and developed AI chips, and it has acquired a wholly-owned subsidiary Zhongtianwei; Facebook has joined the independent research and development AI chip camp; Jingdong announced the AI ​​panorama, and Zhou Zhihua has joined the AI ​​Institute of Jingdong.
Technology giants have “independently created cores†and bet their own chip R&D, in addition to helping them run better and reduce costs; the most important thing is to reduce their dependence on suppliers.
Giants gather to enter the self-developed chip
No "independently created core" is not a technology giant
On April 20, Alibaba announced the development of an AI chip and wholly acquired Zhongtian Micro. Prior to this, Ali has invested in a number of chip companies such as the Cambrian, Barefoot Networks, Shen Jian, Kneron, and ASR.
Alibaba's acquisition of Zhongtian Micro is only for the purpose of developing a neural network chip, Ali-NPU. It is understood that the chip will be used in imagery video analysis, machine learning and other AI reasoning calculations. According to the design, the chip's price/performance ratio will be 40 times that of current products. The AI ​​chip aims to solve AI reasoning and calculation problems in commercial scenes such as image, video recognition, and cloud computing, thereby improving computational efficiency and reducing costs.
Alibaba acquires Zhongtian Micro
According to industry insiders, Ali's self-developed artificial intelligence chips not only provide long-term driving force for the development of their own businesses, but also have a leading effect for independent research and development of domestic companies' core technologies. It is also a positive attitude to limit the United States.
In addition to Alibaba, Tencent and Baidu have their own layouts for chip R&D. Tencent once led a programmable chip company, Barefoot Networks, which developed the world's first programmable chip capable of processing network packets at 6.5MB/s.
Baidu is deploying autopilot and intelligent voice field. It once led the pre-B round financing of Zhongyuan Huiyan's 100 million yuan. The company is dedicated to autopilot systems and third-generation ASIC chips, etc. In terms of chip R&D, Baidu is mainly engaged with In cooperation with ARM, Violet, and Hanfeng Electronics, DuerOS chips with integrated RDA5981 (violet), mbed (ARM), and HF-LPB200U modules (Hanfeng) were developed.
It can be said that the occurrence of the "ZTE Incident" has further strengthened the determination of domestic BAT companies to invest in self-developed chips. In addition, it also serves as a warning to some Internet companies abroad. Facebook also announced its own research and development of chips. According to sources, Facebook is building its own chip design team. According to a person in the circle of science and technology, Facebook designed its own chips to reduce its dependence on chip giants such as Intel and Qualcomm.
Facebook self-developed chip
In addition to Facebook, Apple also began developing its own chips as early as 2010, mainly because Apple wanted to get rid of its baseband chips for Qualcomm and its dependence on Intel's processors. It is understood that an iPhone needs to pay $40 in royalties to Qualcomm. In 2016 alone, Apple paid patent fees to Qualcomm up to $8 billion. Currently, A-series processors on iOS devices and W1 chips on AirPods are independently developed by Apple.
Google is also actively developing its own processor. On GooglePixel2, which was launched last year, it has used a self-developed chip, but this chip is designed to handle heavy image operations. Of course, Google also Independently developed the AI ​​chip - TPU.
Why technology giants independently research and develop chips
BAT is also good, Huawei is also good; the United States of Google, Facebook, Apple, etc. have invested huge amounts of money and manpower to engage in self-development chip, mainly due to the following three considerations:
The first point is that they want to break the technology blockade of chip makers and increase their voice. Chip makers are absolute and unreliable upstream supply chain vendors for mobile phones and Internet companies. This determines what products the upstream company gives you and what products you need to digest. For example, at the beginning of 2015, Qualcomm released the Xiaolong 810 chip because of severe heat, dragging down all domestic mobile phone manufacturers using this chip.
In addition, a large number of patent licensing fees are involved, causing many manufacturers to suffer. If the patent fee is not negotiated, upstream chip makers will cut off supply. For example, Meizu did not negotiate patent fees with Qualcomm earlier, resulting in Meizu’s flagship phones that can only use MTK chips or Samsung Orion processors.
The second point is to improve the industrial chain to save costs. Take the mobile phone as an example, the core components of the current mobile phone are imported products. As long as the cost rises, the downstream hand manufacturers will have to follow the price increase. The success of self-developed chips will reduce the reliance on the upstream manufacturers and improve their bargaining power. The most important thing is that since it is developed by oneself, this can shorten the development time in the later product experience optimization.
The third point is that the "self-research chip" itself has a halo, which will increase brand awareness to some extent. The introduction of self-developed chips not only means that manufacturers have made great technological progress, but also will greatly increase the brand value of manufacturers and increase consumer trust. For example, Huawei's HiSilicon processor has successfully opened its reputation in the market.
Self-developed chips will not happen overnight
The independent research and development of the chip must first clarify a point here. That is, independent research and development does not mean that they “start from scratch.†The popular point is that what we need to do now is to re-do what others have already developed. One of the nouns here is certainly familiar with the "localization rate."
Take high-speed rail, for example, in the early stage of high-speed rail construction (2004), almost all domestic technologies and patents for Germany and Japan were introduced, including the core IGBT chips. Until 2015, China realized the mass production of IGBT chips. The real application of this chip was actually a year later, in 2016. Compared to earlier imports of high-speed trains, the localization rate of high-speed railways has been increased. 90%.
From the perspective of the above development, the chip is a long-term investment product. The 12-year period from the introduction of the high-speed rail chip to the official mass production business will enter the payback period. This requires the establishment of a sound intellectual property protection system in addition to R&D, as well as institutional and financial support. This requires national guarantees.
In addition, for the time being, we want to acquire core technologies that are either R&D or purchased. According to the high-speed rail case, we can see that the market-oriented solution is in line with the current environment.
Kindwin Technology (H.K.) Limited , https://www.ktlleds.com