The general trend of the world will be united for a long time, and it will be divided for a long time.
At the end of the week, the Seven Kingdoms were divided and merged into Qin. Qin II perished, and Chu and Han were divided and merged into Han. Han rebelled since Gaozu beheaded the white snake, unified the world, and later Guangwu Zhongxing spread to Emperor Xian, and was divided into three kingdoms. "Long time will be united, long time will be divided" comes from here.
Recently, there have been many major events in the scientific and technological circles, which has the meaning of "seven countries competing for dominance and three countries competing for hegemony". Recently, Nvidia, AMD, Intel (Intel) three chip giants have announced the results of the acquisition, and Nvidia also released a new financial report. Metacosmity is regarded as an important application scenario of the next generation of technology, as the three chip giants, how will it be laid out on the technology route? We may be able to learn a thing or two from the relevant acquisitions and financial reports.
Nvidia's earnings report exceeded expectations
On February 16, local time, Nvidia, which is regarded as the entrance to meta-universe hardware, announced the company's fourth quarter and full-year financial results for fiscal 2022, as well as the next quarter performance guidance.
Fourth Quarter Results:
NVIDIA's adjusted net profit for the fourth quarter was $3.35 billion, up 71 percent from $1,957 million in the year-ago quarter and up 13 percent from $2,973 million in the prior quarter, earnings reports. It is worth mentioning that NVIDIA's fourth-quarter fiscal 2022 revenue and adjusted earnings per share exceeded wall Street analysts' previous expectations.
Full Fiscal Year 2022 Results:
For the entire fiscal year 2022, Nvidia's net profit was $9.752 billion, up 125% from $4.332 billion in fiscal 2021 and $3.85 per diluted share, up 123% compared to $1.73 in fiscal 2021.
Excluding certain one-time items (not in accordance with U.S. GAAP), Nvidia's fiscal 2022 net profit was $11,259 million, up 79% from $6,277 million in fiscal 2021, and diluted earnings per share were $4.44, up 78% from $2.50 in fiscal 2021, also exceeding analyst expectations.
Outlook for the next quarter:
Nvidia expects the company's revenue to reach $8.1 billion in the first fiscal quarter of 2023, up or down 2 percent, beating analyst expectations, and gross margins expected to be 65.2 percent and 67.0 percent, up and down 50 basis points, on a par with and without U.S. GAAP margins.
Sub-business highlight: Data Centers
Core business performance: The game business, the data center business and the professional visualization business all reached new highs. The game business and the data center business are the most core businesses of the company's financial report, accounting for more than 80% of the two together.
NVIDIA is the world's leading computing chip company, and its products are not only used in traditional fields such as games and data centers, but also in emerging fields such as autonomous driving and meta-universes. As long as there is a need for data calculations, NVIDIA can be seen.
Since last year, the concept of metaversmity has been hot, and a number of technology companies have announced their entry, and the data center, as one of the "infrastructure" of metaversity, is of great importance. Data centers are related to cloud computing and edge computing, and one of the most basic conditions for the realization of the metacosm is the breakthrough of computing power.
According to the financial report, NVIDIA's data center business accounted for a continuous increase, reaching 42.7% in the quarter, and is expected to surpass the game business to become the company's largest business in the future. The data center business has rapidly increased from 20-30% to about 40%.
NVIDIA's data center business achieved revenue of $3,263 million in the fourth quarter of fiscal 2022, and the year-on-year growth rate continued to rebound to 71%. NVIDIA's data center business continues to grow beyond market expectations. The impact of the company's acquisition of Mellanox, which was previously consolidated, ended in the second quarter of 2021. Subsequently, the company's growth rate in the second half of 2021 increased mainly due to the company's endogenous growth and the continued growth of the downstream cloud business (downstream cloud vendors are mainly Amazon).
Data Center Strategy: The Foundation of the Metacosm
The importance of the data center has even gradually affected the strategic level. On the domestic side, on February 17, the National Development and Reform Commission jointly issued documents with multiple departments, and the "East Counting West Calculation" project was officially launched.
"Number" refers to data, "calculation" refers to computing power, in simple terms, eastern number and western calculation refers to the transmission of data from the east to the west for calculation and processing, similar to the south-to-north water diversion and west-to-east power transmission, which is another important strategic project of the mainland.
ACCORDING TO CITIC Securities, the domestic launch of the "East Number West Calculation" project, that is, the data with low latency requirements (mainly cold data such as background processing, offline analysis, storage and backup) is transferred to the western data center, while the data with higher network latency requirements (such as artificial intelligence, video conferencing, etc.) is still calculated in the eastern hub, promoting the rational layout of the data center, load balancing between the east and the west, optimizing the supply and demand pattern, and guiding the data center to develop in the direction of intensification, scale and greening.
At present, when we mention the hardware basis of the metaverse, most of them still consider downstream applications based on devices (such as VR/AR), but data and computing power are actually more basic premises for the realization of the metaverse. The breakthrough of data and computing power can be seen from the successive introduction of domestic policies, and the importance of relevant levels can be seen.
In addition, data centers are also one of the seven major areas of new infrastructure in China, and have long been high-level attention at the strategic level.
Let's go back to NVIDIA. The company's fiscal 2022 results and the next quarter (fiscal first quarter 2023) performance outlook exceeded expectations, but NVIDIA shares fell after hours on the day of the earnings report. This may have something to do with a message.
Acquisition by Nvidia
Nvidia also announced its fiscal fourth quarter 2022 financial results and announced that it terminated the agreement to acquire ARM, a chip design company owned by SoftBank Group, and therefore included operating expenses of $1.36 billion in the first quarter of fiscal 2023. In other words, Nvidia lost $1.36 billion on the termination of its acquisition of ARM. Although the news was announced in early February, Nvidia mentioned it again at this earnings report meeting when it guided the performance of the next fiscal quarter.
Although the acquisition of ARM was unsuccessful, NVIDIA CEO Jen-Hsun Huang still gave good wishes for ARM's future. Huang Renxun said that NVIDIA's future is bright, and NVIDIA will continue to support them with loyal customers in the future.
Huang Also said that although they and ARM are not a company, they will still work closely with ARM, and Son's significant investment in ARM will expand the CPU architecture from customer computing to supercomputing, cloud computing, artificial intelligence and other fields, and he expects ARM's architecture to be the most important CPU architecture in the next decade.
ARM's current owner is SoftBank, and SoftBank CEO Masayoshi Son said that ARM is not only becoming an innovation center for smartphone transformation, but also an innovation center in the fields of cloud computing, automotive, Internet of Things and meta-universe, and has entered the second stage of growth. Son said they will seize the opportunity to start preparing for ARM to go public for financing to drive further development.
Son said ARM is likely to be listed on the NASDAQ stock market, which is dominated by technology stocks, because many of ARM's customers are in Silicon Valley.
Son said, "Our goal is to achieve the largest IPO in the history of semiconductors. ”
For the reasons for the failure of NVIDIA's acquisition of ARM, the domestic brokerage company's capital electronics team boils down to the following points:
1) ARM customers are mostly Nvidia competitors, although Nvidia promised that its acquisition will not change ARM's openness and neutrality, it will still treat all ARM customers equally, but as ARM customers, Qualcomm, Google, Microsoft and other giants are bound to have doubts about this, Qualcomm and Microsoft and other companies have also expressed concerns about this acquisition to the regulatory authorities.
2) The anti-monopoly authorities of various countries have strictly reviewed and no regulatory authorities have approved the acquisition. Under national and regional laws, Nvidia's acquisition of ARM requires the consent of regulatory authorities in the United Kingdom, China, the European Union and the United States, and no regulatory authority has approved the transaction until the two parties have declared the transaction a failure.
ARM itself does not sell CPUs, it provides three different external licensing models according to the different needs and capabilities of chip design companies: one is ARM through the upfront licensing fee for IP licensing, including ARM architecture and IP; the other is to extract royalties proportionally according to the price of each chip; and the third is to benefit from new software services. Although ARM's instruction set is not open source, as long as you spend money, you can get the instruction set license from ARM.
The reason why Nvidia spent a lot of human and financial resources to promote this transaction is that it hopes to push its AI strategy to a new height through GPU and data center business. At last year's GTC conference, NVIDIA launched a virtual avatar platform that can carry the vision of the "metacosm", "Omniverse Avatar", which omniverse has called "the infrastructure of the metacosm". With the improvement of computing power and the enhancement of data processing capabilities, NVIDIA's AI strategy may be able to raise to new heights.
Regarding NVIDIA's report on the last fiscal quarter and the achievements of artificial intelligence and meta-universe, see the author's article last year, "Meta-universe and artificial intelligence, is NVIDIA's "road to advancement"? 》。
Acquisition of AMD
There is no harm without contrast. Nvidia's old rival in GPU (graphics display chip) AMD (Chaowei Semiconductor), but recently announced that it has won Xilinx, the world's largest FPGA manufacturer. According to Xilinx's market capitalization, this acquisition is worth nearly $50 billion ($49.8 billion), setting a new record for acquisitions in the chip field, and this acquisition also successfully refreshed the history of AMD (author's note: In 2006, AMD acquired the graphics display chip manufacturer ATI, and since then, Nvidia's N card and AMD's A card dispute have been "enduring" among domestic computer fans, until Intel inserted a foot in the core graphics field).
Compared with NVIDIA's acquisition of ARM, AMD's acquisition can be described as very low-key, but significant.
At present, the FPGA, which is very popular in the semiconductor industry, is a hardware programmable chip, has better performance and lower power consumption, and is regarded as the "Lego" of the digital world, which can be customized to develop a variety of computing acceleration functions. FPGA chips are the basic products of 5G communications, data centers and many other market hotspots, whether it is edge data processing or high-performance computing in data centers, they can benefit from it.
Although there are many advantages, the development of FPGAs is very difficult. On the one hand, the chip materials and processes are more demanding; on the other hand, the programming language of the FPGA is more low-level, less popular, and more difficult. These two characteristics have also formed a high degree of monopoly in the industry.
In the global market, Xilinx, Altera, Lattice and Microsemi have almost monopolized the FPGA market. Among them, Xilinx is in a leading position in technology, accounting for more than 50% in China and the world.
Founded in 1984, Xilinx, along with Altera, is the inventor of FPGA, programmable SoC and ACAP.
In 2015, Altera was acquired by AMD's old rival Intel for $16.7 billion (in other words, the industry's second oldest had been acquired by Intel, and this time AMD acquired the industry leader). AmD's acquisition of Xilinx means that the two largest companies in the FPGA space have been integrated by semiconductor leaders. The global FPGA competition landscape will also produce new changes.
Obviously, for the industry giants with sufficient funds, taking the road of self-research and development may not be a good choice, so "spending money to buy" has become an optimal solution ("banknote ability" is very important, such as Microsoft's sky-high acquisition of Activision Blizzard some time ago, so as to compete with Sony in the handheld field). And this amD smooth acquisition, can be described as a tiger.com. Nvidia's financial report and future highlights lie in the data center and the related construction of the meta-universe "infrastructure", if AMD also exerts efforts in this field, then the future is obviously "good drama". This will benefit the industry, not a monopoly.
AMD CEO Lisa Su also said that with the acquisition of Xilinx, AMD will expand the market in data centers, 5G communications, automotive, industrial, aerospace and defense. "These are markets where AMD is rarely involved, and they all require high-performance computing."
Attached: AmD's acquisition, the approval of China's regulators played a final role. However, there are restrictive conditions for the approval of the acquisition in China, as detailed in the official website of the regulatory authorities, "Announcement on the Approval of the Anti-monopoly Review Decision of Chaowei Semiconductor Company's Acquisition of Xilinx Company's Equity In the Case of Chaowei Semiconductor Company". Released on January 27.
Acquisition by Intel
In addition, in the chip field, Intel, another strong rival of Nvidia, has also recently been relatively smooth in acquisitions. On February 15, Intel announced plans to acquire Tower Semiconductor, an Israeli semiconductor solutions foundry, for $53 per share, an acquisition expected to cost approximately $5.4 billion. According to Intel's official announcement, the transaction has been approved by the boards of directors of both parties, and is currently going through the relevant transaction process and approval to the regulator, and will be completed within 12 months if there is no accident.
Tower Semiconductor, the world's ninth largest wafer foundry, focuses on analog chip production, and currently has a market capitalization of $3.594 billion, with expertise in RF, power supply, industrial sensors and other professional technologies, and operates foundry business across regions in the mobile, automotive and power supply markets.
At present, only a very small number of companies in the world can independently complete all the processes of chip design, manufacturing and packaging and testing, including Intel, which will greatly improve its industrial chain and increase the competitive chips with TSMC and Samsung.
Although the acquisition has not yet been formally completed, Intel is already looking forward to a bright future. Randhir Thakur, president of Intel Foundry Services, said that the addition of Tower Semiconductor will greatly enhance Intel's competitiveness in the foundry business:
"We are delighted that the Tower team has joined Intel, whose decades of foundry experience, extensive customer resources, and deep technology accumulation will accelerate the development of Intel foundry services."
Behind the large-scale acquisition is Intel's ambition to expand into more chip foundry fields.
At the time of the decline in the original deep layout of the PC market, Intel began to increase the opening of wafer foundries to the outside world, opened its chip foundries to all chip companies, and gradually wanted to turn former competitors such as Qualcomm, Samsung, MediaTek, and NVIDIA into customers of its own wafer foundry business. Last year, Intel launched the IDM 2.0 strategy to accelerate the degree of openness, and the business became a combination of "Intel factory + third-party capacity + foundry services".
Recently, Intel announced the establishment of a $1 billion Foundry Services (IFS) Innovation Fund to make strategic investments in start-ups in the chip foundry field. As part of Intel CEO Pat Kissinger's IDM 2.0 strategy launched last March, it was a clear signal that Intel was once again opening up its foundry services to the outside world.
In an article published by Intel CEO Pat Kissinger on February 11, he said: "With the growth of massive computing power demand, we are entering a new golden age of semiconductors. Intel will continue to lead the "Golden Age of Semiconductors" and contribute to the creation of a globally connected industry. (In the PC era, under the x86 architecture, Intel has "led" the development of the chip industry for nearly half a century, see the following architecture section)
From this point of view, the same is the acquisition, among the three companies, NVIDIA is the most miserable. However, it is worth mentioning that in January, Nvidia announced that it had acquired Bright Computing, a high-performance computing cluster management company, for an undisclosed amount. Unlike its acquisition of ARM, Bright's deal is a simple acquisition aimed at expanding NVIDIA's customer base with new opportunities, and it is not difficult to see Nvidia's desire to grow its business and tell a new story.
Analysis of the technical route: the ambition behind the three acquisitions
In addition to expanding into the field of foundry, Intel has also made a big move in the choice of technology routes.
On February 8, Intel announced its participation in the RISC-V Foundation.
RISC-V International, the global open hardware standards organization, announced that Intel has joined the RISC-V Premium membership queue. Dr David Patterson, co-founder of RISC-V, said: "I am delighted that Intel, the company that pioneered microprocessors (chips) 50 years ago, is now a member of RISC-V International. ”
Behind the acquisition, it reflects the ambitions of the three chip giants and the route of subsequent development:
NVIDIA acquired ARM, trying to hold the ARM architecture in its hands, so as to better develop the digital center business and the field of mobile chips, and if the acquisition results, it can form the layout of the whole industry chain of CPU + GPU + DPU;
AMD's acquisition of Xilinx is to improve its own products and services in the data center sector and make up for the shortcomings;
Intel's joining RISC-V is different from the previous two: he is a "can't do it, you can only choose to join" situation, which belongs to consolidating its industry position.
The actions of these three giants are all rushing to monopolize technology in a certain direction.
Before analyzing the intent behind the acquisition of the three giants, let's clarify two concepts: architecture and instruction set.
A instruction set is a command that instructs a chip to perform some kind of operation or process function.
Instruction is the smallest functional unit of hardware operation, the set of all instructions on a computer, is the instruction system of this computer, the instruction system is also called the instruction set.
The architecture, on the other hand, is a hardware circuit built to implement the instruction set.
Some define architecture as a hardware implementation, while others see architecture as equivalent to instruction sets. A instruction set can design different hardware circuits, and this is the architecture.
A mature and applicable architecture requires years of technical accumulation to be successfully developed, which is also a major problem in the neck of the mainland chip field (money, time, people, indispensable).
The three most critical architectures today are X86, ARM, and RISC-V.
X86 is clearly Intel's ban, and the only companies that have a permanent license for X86 so far are Intel and AMD. Since Intel introduced the Intel 8086 processor in 1978, the x86 architecture has been in the turbulent chip market for more than 40 years (nearly half a century!). ）。 According to a report released by IDC in 2017, x86 processors have a 96% share of the server market. (Small popular science: The personal PC computers we currently use, the CPU is basically an x86 architecture processor. In the CPU chip consumer market, there are only two, Intel and AMD, have you ever seen a third CPU chip manufacturer? ）
The history of the development of X86 chip architecture and how Intel has become so powerful from scratch, and how his competitor AMD has reversed adversity and almost killed Intel, for details of these historical stories, see the author's previous article "Semiconductor Chronicles: The Annihilation and Birth of legends".
As a CPU supplier, AMD, as the "second oldest", has long been subordinate to Intel (see the comparison of all CPU market shares in the figure below, we can see that there is still a certain gap between AMD and Intel). Among the many CPU market segments, AMD achieved a counter-overtaking of Intel in the desktop CPU market in the first quarter of 2021 alone: AMD surpassed Intel, which accounted for 49.2% of the global desktop CPU market, with a share of 50.8% of the global desktop CPU market.
Let's go back to the recent acquisitions of three chip giants.
Nvidia's acquisition of ARM is obviously to directly get its own (in the field of mobile chips) "X86", the CPU chips we currently use in mobile phones are basically based on the ARM architecture, ARM has developed into a standard publisher of public CPU, and no one wants him to fall into the hands of any single company. And this time NVIDIA wants to get ARM, the failure of this acquisition, I am afraid that Intel and AMD competitors have played a lot of resistance.
Wang Zhiwei, a semiconductor analyst at Xintai Securities, said that NVIDIA's acquisition of ARM is aimed at laying out data centers and mobile chips. Prior to this, NVIDIA has completed the GPU+DPU pattern (it is itself a GPU giant), and the acquisition of ARM, a key instruction set architecture, can form the layout of the whole industry chain of CPU+GPU+DPU. Even though a failed acquisition may limit Nvidia's freedom, it will have little impact on its business.
(Trivia: DPU is the abbreviation of Data Processing Unit, which is a large category of the latest development of special processors, and is the third important computing chip in the data center scenario after the CPU and GPU, providing a computing engine for high-bandwidth, low-latency, data-intensive computing scenarios.) ）
Both X86 and ARM are relatively closed, while RISC-V is different because it is "completely open source." For the RISC-V architecture launched in 2010, unlike the x86 and ARM architectures with licensing restrictions, RISC-V has the characteristics of thin, open source, neutrality, and fast response speed, which can develop unique chips that are more suitable for specific needs, and break the conventions of x86 and ARM architectures with high licensing fees and customization difficulties, and are expected to become the next generation of important chip instruction set architectures in the world. Market research firm Tractica noted that the MARKET for IP and software tools for RISC-V is expected to reach $1.07 billion by 2025.
The market of RISC-V is such a large market, and it is open source, which has touched the market position of Intel x86 architecture. Intel last summer considered buying RISC-V-focused chip design startup SiFive for $2 billion, but the acquisition failed. Recently, Intel announced the addition of RISC-V, which is obviously a careful consideration of "joining if you can't fight". After all, RISC-V is such a fast-growing market, as the head player, Intel, it is impossible not to pay attention.
At the same time, Intel's entry into RISC-V is also paving the way for its expansion of foundry. Intel is fully committed to the chip foundry business, aiming to restore its dominance in chip manufacturing, and a more open Intel is a prerequisite for achieving this goal. As mentioned earlier, Intel announced the establishment of a $1 billion foundry services fund to support start-ups and established companies with great potential in x86, ARM, and RISC-V to help companies build disruptive technologies using Intel foundry services.
In addition, Intel has developed an open Chiplet platform that hopes to put together the dies of the x86, ARM, and RISC-V cores and package them into a coherent chip through Chiplet chassis designs.
These changes will have a profound impact on the industry.
Calista Redmond, CEO of RISC-V, said: "The open collaboration of RISC-V has triggered a profound transformation in the semiconductor industry, and this partnership will accelerate innovation in open computing. ”
Bao Yungang, a researcher at the Institute of Computing Technology of the Chinese Academy of Sciences and secretary general of the China Open Instruction Ecology (RISC-V) Alliance, said: "Intel joined the RISC-V International Foundation, and personal judgment will further promote domestic enterprises to increase investment." It is believed that domestic enterprises and institutions will also become the core contributors and leaders in the field of RISC-V like the communication field. ”
The layout of Chinese chip companies in RISC-V is still quite stable, with 19 distinguished members (senior members) of the existing RISC-V foundation and 12 Chinese enterprises (including Taiwan enterprises), occupying the majority. See the chart below for a list of elite members of the RISC-V website (as of February 18, 2022).
RISC-V is not the first open source instruction set, but it is important because it is the first instruction set architecture designed to be able to choose the appropriate instruction set for a specific scenario. Based on the RISC-V instruction set architecture, it is possible to design server CPUs, home appliance CPUs, industrial control CPUs and sensor CPUs.
As an open source instruction set architecture, RISC-V gives users the opportunity to avoid the lock-in of Intel's x86 intellectual property system and ARM's high chip royalties, which makes companies in the global chip industry pay great attention and interest to RISC-V.
Speaking of which, have readers thought of anything? This is the opportunity for domestic substitution and overtaking in curves. This also explains why there are so many Chinese companies among the distinguished members of the RISC-V Foundation. In the absence of a first-mover industry advantage, the open source attribute of RISC-V is very friendly to chip research and development of domestic enterprises.
In terms of domestic policy support, Shanghai has issued the "Notice on Carrying out the Second Batch of Shanghai Software and Integrated Circuit Industry Development Special Fund (Integrated Circuit and Electronic Information Manufacturing Field) Project Declaration Work in 2018", which includes the direction of processor chips based on RISC-V instruction set architecture: support the research and development and industrialization of processor chips based on RISC-V instruction set architecture, 32 bits and above, and the core needs to have independent intellectual property rights.
This is the first policy in China to support RISC-V.
As for AMD's acquisition of Xilinx, in addition to the above, AMD will become a chip manufacturer in the global market that can provide cpu, GPU and FPGA products at the same time. This is critical.
Let's go back to the acquisition targets of the three chip giants:
If Nvidia wins ARM, its monopoly position is equivalent to taking the mobile version of the "x86" architecture, the layout of the CPU field, but also an important part of the layout of the data center, can form the layout of the whole industry chain of CPU + GPU + DPU; Intel's acquisition of high towers, open foundry, and entry into RISC-V is to consolidate its leading position in the CPU field; and AMD's acquisition of Xilinx can also become a manufacturer of three chip products, spanning the three major architectures of x86, ARM, and RISC-V. Apparently, all three companies are trying to achieve some degree of monopoly.
Is there a feeling of "Romance of the Three Kingdoms"? Will the future be "three families returning to the top", or "competing in court"? On the eve of a major technological breakthrough, it's exciting to think about. The innovative breakthrough of chip architecture has not changed significantly for many years, and the breakthrough of computing power is also a delay in major progress, although in terms of materials, the technical route of three generations of semiconductors has indeed brought some changes in the "post-Moore era", but it is still unable to support new and subversive lifestyle application scenarios such as "metaverse". Behind the rapid growth of the chip industry, the industry is in urgent need of major changes. Behind the three giants announcing acquisitions, this may represent the future development direction of technology. For details of the three generations of semiconductors, see the author's article "The Post-Moore Era: The Rise of Three Generations of Semiconductors".
Currently, whether on computers, data centers, or mobile devices, the products of these three American chip companies overlap in a variety of ways. In the future of global sales, the three major companies will balance each other.
High-growth chip industry: the heat is not decreasing
Although the above acquisitions are all American chip manufacturers, China is definitely one of the important players. According to the data released by the SIA, China is still the largest semiconductor market, accounting for about one-third of global sales. China's chip sales totaled $192.5 billion in 2021, up 27.1 percent. From a legal point of view, whether it is NVIDIA or AMD, the relevant approval of the acquisition is inseparable from the approval of the Chinese regulator (and plays one of the key roles).
Although this year is less than 2 months, if there is no accident, the capital heat of the chip industry in 2022 is still high. According to SEMI's forecast, by 2022, the semiconductor industry will achieve three consecutive years of continuous growth and usher in a super cycle.
At the macro level, plans by China, the European Union, Japan, South Korea and the United States will double capital expenditures in the chip industry from $150 billion in 2021. In terms of industry manufacturers, they also expressed their determination at the beginning of the year. TSMC plans to invest up to $44 billion this year in factory building, chip research and development and other activities, Intel directly announced a $20 billion factory, and GF also received $3.2 billion in factory support from suppliers.
At the micro level, a number of overseas semiconductor companies have disclosed their performance a few days ago, and the shortage of semiconductors in 2022 may continue. AMD earnings report shows that fourth-quarter revenue of $4.826 billion, YoY +48.8%, due to cloud computing and enterprise customers increasingly adopt EPYC (Xiaolong) processors, data center revenue doubled year-on-year throughout 2021, with the existing product portfolio of mass production and the launch of the next generation of personal computers, games and data center products, the company expects to continue to usher in significant growth in 2022.
Qualcomm disclosed performance shows that the revenue in the past quarter was 10.705 billion US dollars, YoY +30.0%, of which mobile phone chips achieved revenue of 6 billion US dollars, an increase of 42% year-on-year; Internet of Things (IoT) chips achieved sales of $1.5 billion, an increase of more than 40%; RF front-end chip and automotive chip revenue was $1.1 billion and $256 million, respectively.
The company believes it is currently at the beginning of one of the biggest opportunities in history, and over the next 10 years, the company's potential market will expand more than 700 billion to about $700 billion. Meanwhile, Infineon, a leading supplier of microchips to the automotive industry, expects its 2022 revenue target of about $14.7 billion and believes that the global semiconductor shortage will continue this year.
In the case of the acceleration of mergers and acquisitions and expansion of head players and the continuous background of "lack of core", can the chip industry in 2020 break through the previous growth to a new level? Worth looking forward to. What is more noteworthy is that the three giants of chips have accelerated the layout of future science and technology, and in the choice of major scientific and technological development routes, the three giants have their own emphases, and the metacosm seems to be getting closer and closer to us.
Xu Yuanyao