The leader looks at the chip and walks in front of the world
The leader looks at the chip and walks in front of the world
In recent years, "crossover core making" has already become a hot topic in the science and technology industry.
Whether it is car companies, mobile phone manufacturers, Internet companies or home appliance giants, one after another into the "core building" track, "self-research chip" seems to have become a big trend that can not be blocked.
In fact, as early as 2010, Apple released iPhone4 clearly announced to the outside world that the self-developed processor A4, "core" is no longer the exclusive of chip companies, Google, Tesla, Microsoft, Amazon, Baidu, Alibaba and other enterprises have successively crossed into the market.
Especially in recent years, the increasing user demand, and the semiconductor industry has experienced "chip shortage" and "technology sanctions" under the background, "cross-border core building" has become a trend.
01 Giant cross-border core building, have poured in
In August last year, China's fast company launched * self-research cloud intelligent video processing SoC chip SL200, officially announced the toB market; at the same time, its old rival ByteDance (TIKTOK headquarters) also strides forward in the chip field, disclosed research chip division server chip, AI chip, video cloud chip three latest progress, also revealed the industry 3 times the salary dig chip talent.
In the mobile phone market, mobile phone manufacturers are particularly busy in the core track, China's four major domestic mobile phone brands "Huami OV" gathered, Xiaomi launched ISP chip surging C1 and charging chip surging P1; vivo announced self-developed ISP chip V1 / V2; OPPO launched * self-developed image dedicated NPU chip Mariana X, it is revealed that OPPO has more chips on the road.
Mobile phone factories cut into the field of chip research and development are basically based on self-research, and mostly focus on the improvement of image ability. Among them, although OPPO's Mariana MariSilicon X is an NPU chip, it also integrates ISP and is an NPU of ISP + AI accelerator.
Compared with the mobile phone SoC chip of the master, ISP that only focuses on images is relatively not so complex in technology. More importantly, when the image has become the core competitiveness of mobile phones, how to effectively improve the image performance has become the main direction of "inner volume" of mobile phone manufacturers.
On the other side, the car company building core is not a new topic.
After the lack of core, the impact of giant self-developed chips and the change of automotive architecture on the new demand for semiconductors, car companies began to feel the need to design chips or participate in chip design.
As a local representative manufacturer of the new power of car manufacturing, It carries the banner of "self-research chip".
As early as 2020, it was reported that Nio and Xiaopeng, two new forces, were rumored to build their own autonomous driving (AD) chip development team. As of last November, nio AD chip research and development team scale has reached 500 people, and the research and development is progressing smoothly. It is also revealed that Nio is developing AD chips and lidar chips at the same time; Xiaopeng has chosen Tesla FSD chips as its target product.
In May last year, Ideal Intelligence, a wholly owned company, was formally established, which seems to announce the ideal's determination to enter the chip industry. Earlier this year, Ideal appeared to be working on autonomous driving chips, with its website and third-party recruitment platform looking for SoC system architect and senior engineer in self-party MCU development.
At the same time, the new forces are also trying to attract the industry talent, to promote chip research and development work. The industry expects that Nio, Xiaopeng and Ideal autonomous driving chips may get on the bus around 2024 at the earliest.
From talent grab to investment, the chip battle for new car makers has entered a white-hot stage, but there is still some way from the light at the end of the tunnel. With Tesla as the coordinate, "Wei Xiaoli" still has a lot of work to make up for.
Beyond addition, Chinese automakers such as Geely, Zero Run and BYD are also developing self-driving chips and power semiconductor devices.
Considering the long cycle of chip research and development, mass production and the high technical threshold, a continuous and stable business pace is needed. And the car company core building, is a vertical expansion of the business, the more important main business of the car companies is to build cars and sell cars. These attempts and efforts depend on how many resources the car company has, the patience to manage and execute, and how many cars they can sell and how long they can live.
In addition, capital cooperation has also become another path chosen by car companies. Starting from 2021, Chinese auto companies have also begun to invest in the chip industry on a large scale to lay out their own automotive chip supply chain.
The advantage of cross-border cooperation with semiconductor enterprises is that domestic car standard chips get more opportunities for car certification. Vertical capital cooperation with industrial chain enterprises is almost one of the choices made by domestic automobile enterprises to improve their own chip industry chain and supply chain.
Home appliance giant's self-research chip layout, and its business is inseparable. According to incomplete statistics, midea, gree, hisense, haier, galanz, TCL, skyworth, konka, changhong and other well-known home appliance brands have been through chip department, a subsidiary or investment chip gen enterprise, actively in the field of chip semiconductor layout, mainly around the MCU master chip, power management chip, connection chip, driver chip and image processing chip and other types of electrical appliance chip layout.
Young people eager to try, the veteran has already established a solid territory.
Since 2018, when Baidu took the lead in releasing Kunlun 1 chip, Alibaba, Tencent and other old Internet giants have successively entered the chip field. Alibaba established a chip company, Pingtou Brother Semiconductor, and Tencent adopted the form of "investment + self-research" to enter the market.
In the ups and downs of the following era, BAT has successively removed a number of business lines, but still firmly held the chip battlefield.
Baidu released the domestic * cloud AI chip Kunlun 1, opened the Internet factory core first. For now, Baidu is fully laying out autonomous driving. The year the Kunlun 2 chip was released last year, it was determined to be used in autonomous driving, intelligent transportation assistant and other fields.
Up to now, Ali Flat Head Brother has formed four series of Xuaniron, Yitian, Hanlight, and Badminton Group, building external ecology externally and serving cloud computing internally.
At the end of last year, Tencent for the first time disclosed the development progress of —— three chips "Purple xiao", "Canghai" and "Xuan" are special chips, respectively used for AI computing, video processing and high-performance intelligent network.
In the core of this matter, Huawei Hisi is a veteran. In its product array, in addition to the Kirin series, which is known for smart phones, there are also baseband chips represented by Balon series, communication base station chips represented by Tiangang series, AI chips represented by Shengteng series, and PC and server chips represented by Kunpeng series. Until later by the United States sanctions, everyone is more clear.
On the whole, all walks of life giants have aimed at the chip track, there is a "not core behind" illusion. But throughout the market environment and trend, although the "core road", but do not do more like "boiled frog in warm water", in the later feeling was hit by the crisis.
02 giant build core, go to a new stage
BAT, Huawei are already familiar with the other; semiconductor industry observation in the previous article... This is not described in too much detail.
In addition to these, the giant core building is also ushering in a new stage.
Hisense chip company, plans to spin-off listing
A few days ago, Hisense Video released to spin off its subsidiary Qingdao Xinxin Microelectronics Technology Co., Ltd. to the technology innovation version of the listing plan.
It is understood that Hisense invested 500 million yuan in 2005 to establish Qingdao Hisense Xinxin Technology Co., Ltd. In the same year, China's * digital video processing chip "Xinxin HiView" VPE 1 X with independent intellectual property rights, which reduced the price of similar imported chips from $13 to $5.
In 2017, Hisense acquired Japan's Toshiba TV to integrate its picture quality chip design team. In June 2019, the chip department Hisense Xinxin Company and Shanghai Hongyou Company were integrated to establish Qingdao Xinxin Microelectronics Company. So far, its picture quality processing chip has been iterated four times. It can be seen that Hisense's chip road mainly revolves around the optimization of the display effect, and has achieved a number of "national first", including the launch of 4K 120HZ ultra HD picture quality engine chip in 2015, the release of fully self-developed 8 KAI picture quality chip in January 2022 and so on.
According to public information, Xinxin Micro is a Fabless mode chip design company focusing on display chip and AIoT intelligent control chip. It is committed to providing display chip solutions for all kinds of display panels and display terminals, and providing frequency conversion control and master control solutions for smart home appliances.
Hisense Video said that through the spin-off, Xinxin will achieve independent listing, and through listing financing to enhance the financial strength, enhance the sustainable profitability and core competitiveness of the enterprise.
On this basis, Hisense video also said in the annual report, the company will continue to expand the semiconductor layout, vertical integration to optimize the display industry at the same time, accelerate the horizontal expansion efficiency of chip products.
Byd semiconductor: stick to the listing plan
As early as 2013, Elon Musk proposed to develop autonomous driving chips. Due to the lack of technology and talent pool, Tesla had to cooperate with Mobileye in the early stage, but its products did not meet expectations, only reaching the L2 level. It took five years for Tesla to reassemble the autonomous driving chip, to 2019, when Tesla officially launched its self-developed AP chip Autopilot HW3.0.
Similarly, BYD Semiconductor, a leading company in the domestic automotive chip field, has also experienced quite a long road of technology cultivation and vehicle verification.
In terms of the market and capital, compared with its electric vehicle sales, BYD subsidiary BYD Semiconductor has seen many twists and turns. After many "suspension", the bell rang again in November last year.
However, at BYD's earnings meeting on March 29, Chairman Wang Chuanfu once again said that "BYD's semiconductor listing plan remains unchanged, but there are some adjustments in the process."
In fact, the listing of the capital to invest in power semiconductors, this is one of BYD's plans. According to BYD semiconductor's previous plan, the company will raise 2.686 billion yuan listed, of which 312 million yuan will be used for the new power semiconductor chip industrialization and upgrade project, 2.074 billion yuan used for power semiconductor and intelligent control device research and development and industrialization projects, 300 million yuan used to supplement the working capital.
In addition, in the face of the continuous growth of the new energy vehicle industry, the new wafer capacity is still far from meeting the downstream demand. In order to improve the production capacity supply capacity and independent control capacity as soon as possible, BYD Semiconductor plans to seize the time window to carry out large-scale wafer production capacity investment and construction. In order to expand wafer production capacity, BYD semiconductor listing period also invested about 4.9 billion yuan to implement the Jinan power semiconductor production capacity construction project. At present, the project has been completed and put into operation, and the production capacity is climbing smoothly. It is expected to reach full production in March 2023, when the production capacity will reach 30,000 pieces / month, which is expected to have a great impact on the future assets and business structure of BYD Semiconductor.
In addition, at the end of the listing, BYD Semiconductor also took over the Chengdu Unigroup project, which is also interpreted by the industry as a step to expand wafer capacity. By optimizing the layout of the company's semiconductor sector, expanding production capacity to take the lead to meet domestic demand, so as to resupply for the whole industrial chain, reduce the proportion of related transactions with the parent company, and finally achieve the purpose of listing.
As a semiconductor supplier, BYD Semiconductor has now achieved mass production of IGBT, SiC devices, IPM, MCU, CMOS image sensor, electromagnetic sensor, LED light source and display products. Among them, BYD Semiconductor is best at IGBT automobile power module, which started self-research as early as 2005, and now has the operation capacity of IDM mode of the whole industrial chain.
At present, in the field of IGBT module, BYD semiconductor has also stood in the domestic head position. Data show that in 2018, BYD Semiconductor released the benchmark IGBT 4.0 technology in the field of vehicle regulation level; in 2021, the IGBT 5.0 technology based on high-density Trench FS achieved mass production. At present, in the field of IGBT module, BYD semiconductor has also stood in the domestic head position.
According to the prospectus, in the IGBT field, BYD Semiconductor ranked second in the world among new energy passenger vehicle motor drive manufacturers in 2019 and 2020 for two consecutive years, and domestic manufacturers, with a market share of 19%, second only to Infineon. In addition, according to caitong Securities research report, in the first three quarters of 2022, BYD semiconductor power module installed market share reached 21.1%, close to the track leader Infineon 25.7% market share.
03 chip self-research or outsourcing, still "make is better than buy"?
China has a long history of debate about buying core chips or doing its own research.
The chip industry thinking of "making is better than buying" has led to the fact that the gap between the domestic semiconductor industry is gradually widened. Especially with the occurrence of ZTE Huawei incident, the discussion of domestic manufacturers on self-developed chip research or outsourcing gradually began to shift. It is not only the local chip enterprises in increase self-research efforts and actively enhance competitiveness. Terminal giants also began to make force to the upstream of the industrial chain, "cross-border core making" quietly rising.
For the above giants, chip making needs to have specific scenarios, not for the sake of doing, but based on the internal needs of short video, video quality, cloud computing or autonomous driving. Of course, no matter what the scenarios, the giant core is mostly for two purposes: internal service for its main business, "cost performance" is the source of the core, at the same time to achieve their own differentiated competitiveness in the same track; external market space to find new growth points.
Fit their own business needs, enhance competitiveness
The former is the most obvious trend for industry manufacturers.
From the perspective of international manufacturers, as early as 2013, Google began to develop TPU chips for AI scenarios, to solve the increasingly huge computing needs and cost problems within the company. Amazon also launched its Nitro1 chip in 2013, also serving its own business. Today, Amazon has three product lines of network chips, server chips and AI machine learning chips, with nine cores in eight years. Amazon has managed to replace Nvidia's chips by developing its own chips to process Alixa's voice assistant computing, reducing costs by 30%. Google also released Argos VCU, its own video processing chip, which replaced tens of millions of Intel CPUs, saving Google 20 billion yuan in capital expenditure.
Amazon, Microsoft, Google and several other technology giants have mastered most of the world's server computing power.This huge computing power demand is enough to motivate them to develop their own server chips. Therefore, the cost performance is the * power of the big factory core.
This core heat to the domestic, the Internet factory have joined the group chat.
For example, Baidu has two major chips, Kunlun and Honghu. Kunlun, as an AI chip, can not only use cloud requirements such as deep learning algorithms, but also adapt to the computing requirements of detailed terminal scenarios such as natural speech disposal, large-scale voice recognition, autonomous driving, and large-scale introduction. Honghu is a far-field voice interaction chip, mainly used in on-board voice interaction, smart home and other scenarios, are coordinated with Baidu's business.
In the same way, Ali's chip also directly serves the ecological construction of Ali Cloud. In addition, it also launched its own cloud server, Shenlong server, and designed its own smart network card chip X-Dragon. In short, by constantly enriching the ecology to enhance its cooperation ability.
Tencent's "Canghai", "Zixiao" and "Xuanling", respectively, for video processing acceleration, AI chip, intelligent network card chip, etc., without exception, pointing to domestic demand.
It can also be seen from the core building plan of Kuaishou and Byte that, with the rapid expansion of the business of Internet companies, the hardware requirements of the upper video business need to start from the actual demand and make in line with their own business characteristics ASIC in the chip definition stage. This is also what the byte leader said that "the company cannot find a supplier that can meet its requirements."
If the existing chips in the market are used, it is bound to make a choice among the mature chips of the established giants, which will undoubtedly increase the investment in hardware and the risk of subsequent business stability. And the general chips on the market are often not as good as the dedicated chips in optimizing the video experience. In this case, the self-developed chip becomes a must option.
On the other hand, generic products provided by Qualcomm, AMD and Intel are increasingly difficult to meet the real needs of Internet technology manufacturers. At the same time, it has led to the rising cost of purchasing chips for many technology companies, due to the weak voice; but now with the release of their own chips, it has some confidence against the big companies. Under a variety of factors, the self-developed chip has gradually become a necessary option.
It is not hard to understand why home appliance giants have developed their own chips. On the one hand, under the trend of national environment for chip independent control, we do not want the lifeblood of foreign investors, so it is determined to develop the chip to improve its core competitiveness; on the other hand, the value of independent research and development for terminal enterprises is to improve the differentiation of products, better hardware and software adaptation and performance optimization.
In terms of creating differentiation advantages, although OPPO, vivo, Xiaomi and other mobile phone companies will not face the dilemma of being stuck, there must be the convergence problem of chips. In order to generate competition in technological power, self-developed chips are also necessary options.
However, for the necessity of the car companies to develop chips, the industry varies.
Some people point out that with the rapid growth of the penetration rate of electric vehicles, the supply chain of smart electric vehicles is also relatively mature. Upstream, nvidia and Qualcomm can provide most automakers with autonomous driving and cockpit chips. Compared with most car companies, it has more scale advantages, can split more research and development costs with huge shipments, and can get lower prices for a single chip when cooperating with OEM such as TSMC.
And the car enterprise own research and development of large computing power chip, is the opposite of the above scale advantage. Car companies need to bear extremely high research and development costs. If all the chips are used for personal use, spread to each car, self-developed chips are not necessarily cheaper than external acquisition.
However, before the arrival of whether the core is necessary or not, and whether the cost is the right problem, the resources and ambitious head car companies will still try to develop their own intelligent cockpit, autonomous driving and other large computing power chips.
Because the chip and the intelligent experience relying on the chip, increasingly become one of the competitiveness of the car companies to sell cars. Autonomous driving and intelligent cockpit chip support the software system upward, and data is transferred in the software. Data is the necessary content for automobile enterprises to perceive user behavior and iterate the system. It is impossible for those who want to continuously upgrade their hardware and software systems, with extraordinary experience and brand stickiness, not to try their own chips and operating systems.
Therefore, car companies including Wei oli have this ambition.On the one hand, the independent control of high-end chips has been driven by policies, and it is a trend and inevitable to support the local semiconductor industry. On the other hand, self-developed AD chips can not only help auto companies to build full-stack self-research capabilities, but also enhance brand value to a certain extent.
Market environment, stable supply chain security
Trade frictions between China and the United States have led to supply difficulties for overseas manufacturers, and accelerated the "domestic substitution" boom in the semiconductor industry.
As mentioned above, "core making" is worth trying for the overall environment of the industry, both from the strategic level.
The advantage is that the development of dedicated chip design is relatively easier for large manufacturers compared to the high threshold of universal chips. And it can improve its core advantages of product experience and service differentiation.
In addition, independent research and development can make it achieve * in cost and process, improve security and flexibility, and make faster innovation at different levels in various stages of long, medium, and short, so that it can take the initiative in chip project approval, progress and delivery.
Expand the market space, looking for new growth points
Finally, we will talk around the "giant core or will explore the market space to find a new growth point" this perspective.
The development of the semiconductor industry over the past few decades is closely related to the trend of the hardware market.
Abroad, Intel's rise has been accompanied by the boom of the PC industry, Qualcomm's development depends on the expansion of smartphones, and Nvidia has taken advantage of games, cryptocurrencies and AI.
Therefore, for the current domestic giants, the most important thing is to find an application scenario of their own, whether it is cloud computing, AI training, or MCU, whether it is competitive, whether it can play, all need to try in the market.
Take Huawei Hisi as an example, its * order comes from the security monitoring market of Dahua Shares, and then relies on the base station and base band chip, and then the well-known Kirin series, relying on Huawei smart phones to thousands of consumers.
But from the progress of the rising show, whether it is Baidu or Ali, or other cross-border giants, it is not easy to open the external market. And from the big factory split out, independent financing, may be the only way to its chip business marketization.
On the one hand, the spin-off listing can broaden the financing channels, reduce the asset-liability ratio, and optimize the capital structure and introduce market-oriented incentive mechanism; On the other hand, after the internal business, it can be more convenient to supply other peers and supply customers in more fields, thus rapidly expanding the market share and finding new growth points.
04 is written at the end
From the current big core of the incoming process, now order the giant made core "export" mostly precisely to their own business, and emerging signs of spin-off seek independent listing and case, and from the "core" it itself and the competitive environment, "core" high threshold determines the incoming "core" factory, it is difficult to avoid the subsequent process of protracted war.
But anyway, the chip industry's "build better than buy" logic is becoming the past, left behind in the rut of history.