At my book launch workshop get together a few days ago, lot of people concerned about the prospects of Intel (ticker: INTC). This is the second post of my opinions on Intel. Please go back and read the first post I wrote a few days ago. Article “How does Intel make money? and the benefits to invest in it“, in order to understand the content of this article.
Loss of market share
After AMD (ticker: AMD) has continued to conquer the market for nearly five years, it has threatened Intel’s leading position and caused an impact on its market share. You can also regard Intel’s decline in recent years as a substantial monopoly from five years ago, but now it has evolved into a duopoly. For optimists, AMD must maintain the market growth of the past five years, otherwise it will be difficult for the duopoly to achieve sustainability. For pessimists, if AMD has maintained the market growth of the past five years, duopoly will become a reality.
In the third quarter of 2021, according to Mercury Research’s market survey, x86 CPU market share: AMD has reached 24.5% (the best year since 2007), and Intel has fallen to 75%, other vendors such as VIA accounted for 0.5% — Please note that this is only a market share survey of x86 architecture, and non-x86 architecture CPUs are not included. So, for example, Apple, Amazon, Qualcomm, Ampere, Alibaba, Samsung, Tencent, Huawei, Xiaomi, SiFive, MediaTek, and in the future Microsoft, Alphabet and other non-AMD, or non-VIA vendors designed with ARM or RISC-V architecture CPUs, and thoses have gradually increased exposure of various personal computers, laptops, and servers are all excluded in this survey.
Competitors on the PC client side come together
This is the Client Computing (CCG) department in my previous post “How does Intel make money? and the benefits to invest in it” , which is the personal computers and laptops you and I use every day.
But it is the non-x86 CPU that makes Intel’s position uneasy. Qualcomm (ticker: QCOM) and MediaTek have already joined ARM-based CPU notebooks. Even Microsoft (ticker: MSFT) and Alphabet (ticker:s: GOOGL and GOOG) have reported to develop their own CPUs for use on Windows and Chromebooks respectively. Apple started to abandon Intel this year and introduced its own ARM-based M1 CPU, which allowed Intel to immediately reduce 10% of the global PC market and losses in the form of billions of dollars.
It can only be said that for the personal computer, Intel will not be able to have a monopoly in the future as before. This is certain. The difference is only how much market will be taken away.
Loss of opportunity on the server side
Intel’s central processing unit (CPU) market share in data centers, where server industry shipment volumes are much lower than personal computer clients, however, margin is much higher than that of personal computer clients. Intel’s highest market share was an unimaginable 99 %. According to Mercury Research’s market survey in the third quarter of 2021, AMD has already captured a staggering 10% of the X86 server CPU market.
However, Intel has missed the rise of artificial intelligence, which is far inferior to nVidia (ticker: NVDA) in graphics processing unit (GPU) advantage. It has almost lost its armor and has no ability to parry. A while ago, in order to resist nVidia’s progress in the cloud data center, the data center department was reorganized (the department with the highest market share and profit of Intel, and the department most threatened by nVidia), and launched a new IPU organization to face nVidia.
AMD is taking advantage of its success in personal computers to gradually erode Intel’s traditional central processing unit market in data centers, and has become the main server central processing unit provider for Alphabet’s data center.
In addition to nVidia and AMD; Amazon, Qualcomm, Ampere, Alibaba, Tencent, Baidu have all launched their own non-x86 server CPUs. Amazon, Tencent, Baidu and Alibaba are currently emphasizing that they are for their own use, but they do not rule out selling them to other companies.
In other departments, Mobileys, which was bought with a lot of money, is more important. The self-driving car chip manufacturer, which spent the highest amount in Intel’s history, was originally a little competitive in the industry. However, in recent years, due to the fact that Mobileyes technology has not kept up with market changes and customer needs, technology and information have been updated very slowly, and many important large-scale customers have been gradually lost. The prospects are not optimistic. In addition to Nvidia’s Orin, Qualcomm also launched Snapdragon Ride, all of which are grabbing the market for Mobileye’s latest EyeQ5. What’s more unfavorable is that companies such as Tesla have already launched or are developing their own chips, and performance is not worse than Mobileyes.
Programmable solutions group
And Altera, the largest FPGA manufacturer acquired in 2015, has become the current programmable group (PSG) department. However, in the past five years after this acquisition, there has not been a synergy effect as claimed at the beginning, and Intel’s market share in FPGAs has also declined slightly.
Intel’s other business has no contribution
Except for the personal computer and server data center, the other departments actually do not contribute much to Intel’s revenue or net profit, that is, investors can almost ignore it. The other departments actually do not have strong competitiveness. This can also be explained in the table above. The proportion of revenue is very small, and some are still losing money.
Last week, Intel announced that it originally planned to acquire SiFive, which is a RISC-based IC design house, for US$ 2 billion, was cancelled.
The process gap
To apply the consensus of Wall Street and industry professionals, “First fill in the process gap between the two generations that lag behind TSMC (ticker: TSM), no just talking.” That is, Intel must do more and say less. Investors are tired the soap operas of Intel’s promise repeatedly.
This gap between two semiconductor manufacturing process generations, “even the best scenario, if everything goes smoothly” cannot be achieved without 4 to 5 years-this is why all technology stocks have skyrocketed in 2020, this is the main reason why Intel is the only big tech stock prices falling more than 17%.
In less than three years, three CEOs have been changed. Intel has only had eight CEOs since its establishment 54 years ago. Three of them have been replaced in the past three years. Everyone knows how serious Intel’s internal situation is. Regardless of the reason, no matter the size of the company, there will be no exceptions. Changing three CEOs within three years is the company’s most serious crisis. Nothing else matters. Had it not been for Intel’s special status and the backing of the US government (this is another topic), the stock price would have fallen into a disproportionate way, and it would have been impossible to maintain it at the current level for so many years.
In order to establish two new plants in Arizona, Intel has requested astronomical subsidies from the US government. The new CEO Patrick Gelsinger also mobilized various forces to lobby the United States to block subsidies for TSMC and further subsidies for itself.
Even Intel’s decision to set up factories in Europe has been asking local governments for money and investment subsidies. TSMC also does this kind of thing, but it rarely talks about it publicly, and it does it all under the table. But Intel is openly talking in front of the media, which has a threat to the local countries (including the United States).
Customers began to develop their own chips
Due to the many factors listed above, Intel’s customers began to develop their own chips. This includes:
- Client PC:
- Apple can’t stand the repeated delays of Intel’s product schedule and does not want to be controlled by others. Coupled with the success of its own Ax series processors for iPhone and iPad, as well as the consideration of profit costs, this year has begun to launch its Mac computers using its own M1 series processors.
- Microsoft has repeatedly introduced notebooks that use Qualcomm as a processor. Moreover, Microsoft has also begun to imitate Apple and begin to design its own chips, and in the future it will launch laptops with its own CPU.
- Alphabetic Chromebook laptops: Although Intel processors are now in large quantities, Chromebook laptops using Qualcomm and MediaTek processors have already been shipped. Moreover, following the same path as Apple, Alphabet began to launch Pixel 6.0 mobile phones using its own processor Tensor this year, and will also launch Chromebook laptops using its own processor in the future.
- The central processing unit on the server side:
- Amazon has released its own server chip to the second generation.
- Huawei has launched the “Kunpeng” series of ARM-based server processors long time ago.
- Alibaba has also launched its own server CPU chiplong time ago. This year, it also launched the 5nm process server processor “E-Ten” series based on the ARM architecture.
- There are still a bunch on the way, including all the big customers you want; including Microsoft, Alphabet, Tencent, Xiaomi, etc.
- Internet of Things and embedded system processors:
- In order to delay its dependence on mobile phones, Qualcomm has recently made a lot of gains in the Internet of Things market, which has put a lot of pressure on Intel. Don’t forget that there are many smaller semiconductor vendors in this field. Intel has never negotiated a bargain in this field where price, performance, power consumption, size, function, and design are the starting point.
- nVidia has Nintendo (ticker: NTDOY), and AMD has long-term orders from Microsoft, Sony (ticker: SONY) and Steam deck game consoles, while Intel has been unable to even touch any side for decades.
- Huawei has already launched a processor platform based on the RISC-V architecture.
- As active as Huawei, Alibaba released the RISC-V-based processor “XuanTie 910” as early as 2019.
- Tencent just launched its own-designed server processor Xuanling series a few days ago, as well as the Zixiao series for AI and the Canghai series for video processing.
- Baidu has already launched its own cloud AI chip Kunlun in 2018, and it has now reached the second-generation improved version. And Sparrow series for audio processing.
- There are still a bunch on the way, including all the big customers you want; including Microsoft, Alphabet, Xiaomi, etc.
Among them, because the server processor chip volume is small, but the profit is high, it is Intel’s main source of “net profit”, and Intel’s market share is still very high (Intel has near 100% x86 server processor market share, now dropped to 90%. Please note which is not same as AI or DPU/ IPU server chip).
Why did Intel do it like this?
First of all, all companies have a life cycle. Most of the large listed companies will rarely be able to generate new ideas or generate new revenue sources within the company on their own energy (Organic Growth in English). This is well known, an iron law. Most people have mentioned the law of large numbers, but it is actually not difficult to understand the reasons. It is called “big company disease” in words that most people can understand; while Buffett thinks very thoroughly, and he calls it “Institutional imperative“, the compulsion of the company ───I have mentioned the importance of this matter in many of my previous posts. Whenever we have the opportunity, we will describe this topic worthy of in-depth discussion in a dedicated article in the future. This is what I saw on the surface, and it really is.
For one thing that impresses me deeply, Intel announced to the public that a certain CEO more than ten years ago vigorously publicly announced that the company’s annual revenue will soon exceed US$ 100 billion. More than a decade later, in 2020, the company annual revenue was 778.7 billion U.S. dollars. And almost all the CEOs of listed companies will complain to public that the company’s stock price is undervalued. It is very dangerous to declare the company’s revenue estimates indiscriminately. The point is that it is now proved that it is impossible to achieve. Intel’s annual revenue growth for the past ten years is not far from inflation and GDP growth. This alone may take to court by investors (this kind of thing often happens in US stocks). But Intel has also been seen by investors crystally that its stock price has already paid for it. And almost all the CEOs of listed companies will repeatedly complain about the company’s stock price being undervalued. Intel’s CEOs are no exception — but what investors want to ask is the most fundamental question that does not require profound knowledge: where is the beef?
In my book “The Rules of Super Growth Stocks Investing”, sections 1-5 and 5-4, both mentioned the lifespan of companies listed on the S&P 500 60 years ago. In contrast, it is rapidly declining. In the 1960s, the average survival life in the index was 32 years, but now it is only 24 years. According to the data of the US research organization CB Insights, the average survival time in the S&P 500 index In 1955, the company’s average survival life in the index was 61 years, and in 2015 it was only 17 years, which may be even shorter in the future. Modern corporate technology companies are rarely able to lead the trend for a long time, and it is even more difficult. There are too many reasons for this. It is really rare for Microsoft and Apple to recreate the peak. It is really not easy. What Intel is currently encountering is the “problem piled like a mountain”. It is difficult to see immediate results in the short term (in three to five years). It is a great achievement if get improved in ten years.
The fundamental problem of “Intel customers develop their own chips” is the biggest killer to Intel. It is difficult for Intel to resist this trend, and there are many reasons for it – including:
- Intel itself has repeatedly disappointed its customers: it has been caused by itself over the course of more than a decade, and it cannot blame others. It is possible to pay efforts to prove it, and slowly recover, but it is impossible to return to its past glory. This is certain.
- The second is the lowering of the threshold for chip design: This of course also includes the formation of the division of labor in the semiconductor industry, and chip designers do not necessarily have to produce it themselves. This is an industry trend and cannot go back.
- The third is the help caused by the US embargo: this is my own view. After Trump, the United States and allies jointly imposed an embargo on China, fearing the rise of China, and then using tariffs, laws, embargoes and other means to destroy global semiconductors supply chain. In the long run, this artificial method will not work at all. Coupled with the pandemic of coronavirus, don’t you see that the revenue and tariff burdens of American companies are the victims, the global semiconductor supply chain is destroyed, the semiconductor production capacity is tight, and China will only be hurt in the short-term. In the long-term, it will not change China’s raising or hurt it. In 2020, China imports about three-quarters of the world’s chips, exceeding US$300 billion. Intel has announced that the revenue of Chinese customers accounts for more than 30% of its revenue, which does not include the chips imported by American companies’ factory produced in China.
Will change the business model of the industry forever
Customers start to develop their own chips. “I personally think this is very serious, because it will change Intel’s business model forever“. Customers will not need to buy yours if they use their own chips, and they don’t need to look at your face. The matter we mentioned has taken place on the server side, client PCs, IoT processors, and embedded system processors.
As long as you go back and read this article, more than two dozen have launched their own CPU chips and become competitors of Intel, and you can know how serious the problem is. This is still well-known and relatively well-known, while ongoing and small ones are not included.
Even if Intel catches up with the gap in the process in the future, it will miss many large customer orders, and the damage will be forever.
For example, in my previous article “How does Intel make money? and the benefits to invest in it” As mentioned at the end of the article, Intel’s worst situation is probably the situation it is now, and it will not go down further. The United States cannot afford to lose it, and it is impossible to die from it. It is not an exaggeration to say that it is too big to fall. Didn’t you all see that the US government began to use all means to help it recover? Including Congressional budgeting support (this is the focus of Western countries’ long-term attacks on China, but isn’t the United States doing the same thing itself?), factory tax reductions, threats to require Asian chip manufacturers to provide operating confidential information, and European allies’s factory support, won chip orders from the US military, and foundry for large customers such as Qualcomm.
What I want to say is that if Intel can really come back to life and return to glory this time, the prospects will be very bright, and it will not be a problem for another 20 to 30 prosperous year. But the question is how long should investors wait for Intel?
- The content of this site is the author’s personal opinions and is for reference only. I am not responsible for the correctness, opinions, and immediacy of the content and information of the article. Readers must make their own judgments.
- I shall not be liable for any damages or other legal liabilities for the direct or indirect losses caused by the readers’ direct or indirect reliance on and reference to the information on this site, or all the responsibilities arising therefrom, as a result of any investment behavior.