Intel is in trouble recently.

First, in WWDC 20 years, Apple announced that it had broken up with its good friend who had cooperated for 15 years and switched to a self-developed chip based on ARM architecture.

Then, from July, NVIDIA's share price soared, replacing Intel and becoming the chip manufacturer with the highest market value in the United States for the first time. Recently, it was reported that NVIDIA would buy ARM.

Huang Renxun, the CEO, was overjoyed when Yingwei became the focus. As early as 2008, he said in an interview, "Intel is the most popular punching bag." Whether it is ridicule or self-confidence, when I say this, I can't help but feel that virtue is out of place and times have changed. Now Lao Huang has pride.

In contrast, Intel seems to have no light on its face. On July 25th, the news that Intel's 7nm chip technology was delayed directly pushed its share price to dive. At the opening of the day, it plunged over 15% in an instant, hitting a four-month low.

Obviously, there is not much time left for Intel. With the tide of the PC era fading, its one-man show may have come to an end. Qualcomm, MediaTek, TSMC and other new stars appeared in black powder, which pushed the old actors to the edge. A group of old IT represented by Intel, IBM and Oracle Bone Inscriptions collectively lost their voices in the mobile era, leaving only regrets.

Gates once bluntly said that leaving a vacuum in the mobile phone market was his "biggest mistake at Microsoft, and it was a completely avoidable technical mistake."

Intel, its long-time ally, also has the same distress, which leads to its hesitation in the mobile field and several fights, but it has always been shy. From mobile phones to tablet computers to baseband construction, what can and should be done has been done, but Intel has never been able to maintain its market position.

Just like a drowning person, the more you struggle, the faster you sink.

The lighting of the New World took place at a dusk in June of 197 1.

Chip designer Jin Fan lit up the small round crystal on the table, just like the electricity in the whole space gathered at the tip of the lightning rod, which opened the prelude to the information age.

This chip, numbered 4004, is the first microprocessor made by Intel. Its birth marked the beginning of the computer revolution, and a huge PC empire set sail from here.

From the initial 4004 and 8008 to the Pentium series and then to the Core dual-core processor, every technological innovation of Intel has injected new vitality into the IT industry, and these great innovations come from Intel's firm adherence to Moore's Law based on the principle of change.

Gordon Moore, one of the founders, once put forward a slogan: "Change is the love of our life. For Intel, this sentence may be as important as Moore's Law itself: the number of transistors that can be accommodated in an integrated circuit will double every two years. Moore's Law is based on technological innovation, which determines that change is the way for Intel to survive and compete.

However, in recent years, Intel has been nicknamed "Toothpaste Factory" because of its slow update speed and small performance improvement, which is increasingly inconsistent with the image of the guardian of Moore's Law.

The period from Core2 to Core i, which is generally missed by science and technology powder, is gone forever. At this stage, the performance of Intel Cup has been greatly improved, and netizens affectionately call it "the era of meat buckle". At the height of the fierce battle between Intel and AMD, Intel processors have changed the war situation with substantial performance improvement, and once again gained absolute dominance in the field of PC chips.

I thought this would be the beginning of a triumphant progress, but I didn't expect it to be the last golden age. After the second generation of Core intelligent processors, Intel's performance growth slowed down.

From Sandy Bridge to Ivy Bridge to Haswell, Intel's performance improvement can only be described as a poor single-digit percentage. In the fifth generation of Broadwell, there was even a retrogression, and "withholding meat" became "squeezing toothpaste".

Insufficient performance improvement, chip update speed is getting slower and slower. The 10nm process introduced by 20 15 was not realized as scheduled, and it was delayed until 20 18. The original plan to achieve mass production on 20 19 also failed, which is really yet we called and urged a thousand times before she started toward us. At the same time, TSMC's 5nm process chip has started to be shipped on a large scale, and it is ready to start the 4nm process this year.

Although due to different standards, the process figures are more of a kind of market promotion, but the strong update speed and publicity of the latter undoubtedly poked Intel's pain point, which is hard to argue.

At this year's CES conference, Intel's current CEO Ke Zaiqi announced that he would mass-produce 10nm chips this year. He said that Intel's "technology is developing at an unprecedented speed, and Moore's Law is the core of acceleration." In the case that Intel's core-making process has not been upgraded for five years, the second half sentence seems unconvincing.

In addition, the wind direction of the market has also changed. Since 20 1 1, global PC shipments have declined for seven consecutive years, directly threatening Intel's core business. Intel's CPU sales are greatly affected. Compared with 20 1 1, the CPU sales of 20 18 decreased by 30%, but its revenue did not decrease because Intel raised the price of a single chip.

The lack of sales and the combination of prices reflect the embarrassing situation that traditional business is stagnant and new business is nowhere to be found.

A series of difficulties faced by Intel are directly related to its failure in the mobile era.

20 19 In July, when Intel announced that it would sell the 5G baseband business to Apple, the money and manpower spent on the mobile attempt for more than ten years went up in smoke.

The old IT, which used to call the wind and rain, was completely caught in the embarrassing situation of the right to speak. When Qualcomm, Samsung and Apple's mobile chips shine brilliantly, Intel, who was too late to get on the bus, left a lonely figure.

In fact, Intel's road to mobile chips began very early.

Back to 1997, Intel acquired the strong ARM architecture from DEC, which was developed by DEC based on the ARM v4 instruction set and has the characteristics of low power consumption.

Since then, Intel has been interested in making up for its shortcomings in low-power processors. As the name implies, at that time, Stronarm was more powerful than the public version of arm architecture.

Around 2000, Intel introduced the Stonearm series processors based on this architecture, which is the earliest mobile embedded chip developed by Intel. But Intel's chip designers obviously can't satisfy the use of ready-made architecture products, so in 2002, the self-developed XScale architecture CPU replaced StongARM.

Compared with ARM processor, XScale has lower power consumption, and it has the hardware acceleration capability of video decoding and 3D rendering as early as 2004. However, the public version of ARM Cortex-A8 architecture didn't have a similar floating-point acceleration unit until six or seven years later, and the performance of XScale was evident at that time.

But unfortunately, Intel failed to stick to this road to the end. In 2006, Intel sold its communication and application processor business including XScale to Marvell for $600 million.

A year before this, Jobs found Otellini, then CEO of Intel, hoping that Intel could provide chips for the original iphone, but he was declined by the other party. In retrospect, this may be the most expensive bad decision in Intel's history. However, judging from the environment at that time, it was difficult for Intel to give up the choice of mobile chips.

2006 is the most embarrassing year for Intel since 1985. AMD, the younger brother of the Millennium, jumped to the top with its high cost performance. At one time, its CPU market share was close to 50%, almost tied with Intel. Three years ago, Intel still occupied more than 80% market share.

Sanders, the founder of AMD, used to be the head of Fairchild's sales department. One year after Noyce and Moore left Fairchild and founded Intel, Sanders also left his old club, rolled up his sleeves and started his own business. For more than forty years, Intel and AMD have been in love and killing each other. Perhaps it is a monopoly of Intel's anti-monopoly investigation, or it may be the same root and the same origin, with an unchangeable doomed love fate. In short, Intel has never wiped out this little brother.

Whenever Intel introduces a new product, AMD can quickly become an excellent competitor with comparable performance and lower price. It is said that when Intel employees finish their work and meet AMD employees, they will always ask, "When will your family produce new products, and we will have jobs after you leave." AMD has been Intel's catfish for more than 40 years.

The change occurred in 2003, when AMD released K8' s Athlon 64 processor, which was ahead of Intel in performance for the first time. On the high-end processor, AMD has released a number of Athlon 64 FX products, which are favored by consumers for their powerful performance.

At the same time, because of the design failure of the third generation P4, Intel suffered from Waterloo, and its CPU sales declined. Until 2006, its market share was tied by AMD.

In addition to the old rival AMD, the ARM camp is also expanding and is gearing up for the PC. Intel's basic disk was in jeopardy, so Otellini made a decision to simplify the administration, cut off the mobile communication business that consumed a lot of manpower and material resources and made little money, and concentrated resources to protect the PC fortress.

In July of the same year, the Core dual-core processing system was born, and Intel once again left AMD behind and regained the throne of computer CPU.

Although it won a great victory in the fierce battle with AMD, Intel also missed the best opportunity to enter the mobile chip market. At that time, the first generation of iPhone was released, and Qualcomm also introduced the first Snapdragon chip, Snapdragon S 1. The smartphone revolution began, but this time Intel failed to become the protagonist.

One year after the launch of Snapdragon chip, in March 2008, Intel also launched Atom series chips for mobile devices. It is eager to rely on this relatively low performance, low power consumption and low price chip to expand the mobile phone market and catch up. I never thought about it, but I bumped into a creative genius-ASUS.

I have to admit that Intel's manufacturing process is really strong, and the Atom chip used in mobile devices can also run on computers. Based on the principle of making the best use of everything, ASUS has developed a netbook, which only has basic functions such as surfing the Internet and watching videos, but it is quite cheap. In 2008, the financial crisis led to the decrease of mass consumption level, and netbooks became all the rage.

Sometimes thinking is completely different from what you really think. Low-profit mobile chips occupy the space of high-profit CPU, and Intel began to have self-doubt.

"Do you want to make a mobile chip?" This issue has been discussed within Intel for four years, and the Atom chip was artificially dystocia for four years. In 20 12, Intel finally made up its mind to make mobile chips, and the market changed.

In this year, Qualcomm Snapdragon, MediaTek MTK, Samsung Orion, Texas Instruments, NVIDIA and other manufacturers rose together, seemingly in their own way, but they were all based on ARM architecture. Looking around, only Intel seems to be out of date, trying to enter the mobile market with X86 architecture.

However, Wintel Alliance vividly learned the lesson of squeezing profits from PC component manufacturers by monopoly. The major mobile manufacturers have extremely rejected Intel in their hearts. Moreover, the ARM architecture has become a recognized standard, and many apps cannot run directly on the closed X86 architecture. At this time, Intel wants to formulate new rules, which is undoubtedly the opposite of the industry ecology with a subversive attitude, which is somewhat out of date.

Only Motorola and Lenovo are willing to give Intel a face. In 20 13, Lenovo introduced K900 with Atom processor, but abandoned Atom in the following K9 10. In the same year, Motorola's Razr i was also equipped with an Intel processor, but Motorola has been forced to go west and can't afford to make waves.

Lenovo and Motorola's attempts ended in failure, and the rest of the manufacturers continued to remain silent and unwilling to cooperate, so Intel had to give up.

However, the failure in smart phones did not stop Intel from entering the mobile market, and the thriving tablet became Intel's next force point.

However, this track is still the territory of Apple, Qualcomm and Samsung, and the unpopular Intel needs to build its own team. Shenzhen Huaqiang North, thousands of miles away from Silicon Valley, provides a stage for this. A large number of white-label manufacturers gathered here, and Intel with deep pockets quickly established cooperation with these manufacturers with huge subsidies.

According to the 20 14 report of Bear Stearns Research Company in the United States, Intel's subsidy for each Atom tablet is as high as 5 1 USD (more than 300 RMB). According to the market situation at that time, the subsidy was equivalent to 1/4 of the price of Android tablet, and it was sold at almost zero gross profit margin. However, the effect of throwing money at the market is really good, and the shipment of tablet computers equipped with Intel processors has reached 46 million units.

Although the shipment results are gratifying, blindly burning money has not brought additional brand effect to Intel. The price advantage of Intel chips disappeared after the subsidy was cancelled, which directly led to a sharp decline in Atom shipments.

Intel, which can't eat meat or drink soup, finally announced in 20 16 that it would stop developing atom series processors and then quit the field of mobile phones and tablets.

But the scale of the mobile market is so huge that it tastes good to chew bones. After the failure of XScale and Atom processors twice, the mobile baseband business became Intel's last chronic disease.

Time is up to 20 10. At that time, the Atom processor was launched for more than a year, and Intel, who was not at ease, made second-hand preparations, that is, it acquired Infineon's wireless business for/kloc-0.4 billion US dollars. Infineon was Apple's baseband chip supplier at that time. Through the acquisition, Intel not only got some orders from Apple and Samsung, but also successfully boarded the 2G/3G through train and went straight to 4G.

In 20 13, Intel released the first 4G baseband product XMM 7 160, which was manufactured by TSMC's 40nm CMDS process. But a year ago, Qualcomm introduced the MDM96 15 chip based on 28nm process, which was adopted by iPhone5. At that time, Qualcomm had big orders such as Apple, Samsung and Xiaomi, while Intel's 4G just started. Backward, but there is still a chance.

For a long time, Qualcomm has used a large number of patents in communication to require major manufacturers to pay taxes and fees at the price of 5% of the whole machine while selling chip packages. For Apple, which takes the luxury route, it needs to pay billions of dollars in taxes to Qualcomm every year. In Cook's view, this tax model is extremely unreasonable. He described: "It's like buying the same sofa, but the salesperson decides the price of the sofa according to the price of the house where each customer lives."

In order to avoid over-reliance on Qualcomm, Apple has always adopted the strategy of Intel-Qualcomm dual baseband supplier, but this balance was broken in 20 17, and the contradiction between Apple and Qualcomm broke out completely, and the two sides launched patent litigation on a global scale. Cook turned the corner and threw an olive branch to Intel, which became Apple's only baseband supplier.

But unfortunately, Intel still failed to seize this rare opportunity.

The problem lies in the technical level. The research and development of baseband chips requires 2/3/4/5G patents. Taking Qualcomm as an example, it has 6,543,800+3,000 patents, covering almost all standards in the whole communication field, and firmly holding the right to speak in the industry.

On the other hand, Intel, through the acquisition of Infineon, set foot in the field of communication, and there are not many patents accumulated. This directly led to Intel's inability in 5G research and development. After waiting for several years, Intel's 5G baseband still hasn't landed, and Apple's 5G mobile phone process can't be delayed any longer, so it has no choice but to throw itself into the arms of Qualcomm again.

20 19 April, Apple and Qualcomm staged a century-long reconciliation, and the patent lawsuit between the two sides, which lasted for several years, finally came to an end. Just one hour after Apple and Qualcomm reached an out-of-court settlement, Intel immediately announced its withdrawal from the baseband business.

There were only two people left in the three-person movie, so Intel had to leave sadly. After selling the 5G baseband service to Apple, this tortuous road to mobileization is completely over.

The chip giant announced a complete failure in the mobile battlefield.

From selling the XScale processor, rejecting Apple, shelving the development of Atom, and finally giving up the baseband business, from today's perspective, Intel seems to have taken a wrong step at every key point of mobileization, but this is a bit wise after the event. Given the circumstances, it's hard to say that Intel didn't make the right choice.

For example, cut off communication services and concentrate resources on pc chips to counter AMD's challenges. After making this decision, Intel's financial performance was excellent, and its annual net profit exceeded $654.38+0 billion, which changed the situation that profits have been declining continuously due to blind expansion.

Intel made the right choice in line with the current interests, but it came to a completely wrong result. Frequent failures in the mobile field are not due to products, but to the distribution of links in the value chain.

Take iOS, Android, and ARM as examples. They all pursue long value chain mode. Their purpose is not only to sell products, but also to take products as the starting point and the node to realize services. Their real core is to take the initiative to accommodate participants and make themselves builders and defenders of ecology, so as to maximize the interests of the group.

On the other hand, Intel pursues a short value chain model. Sell a batch of goods, end an order, and stubbornly insist on doing everything by yourself, and strive to reach out to every link of the industrial chain. It cares too much about suppressing its opponents in order to maintain its monopoly advantage and maximize its personal interests.

Facts have proved that closed monopoly is more than one bargained for in the mobile market. When iOS and Android make money through services and Qualcomm and ARM make money through patents, Intel, which still insists on selling chips, is doomed not to go too far.

In addition, the large plate and the difficulty in turning over are also important reasons for Intel's failure.

Who hasn't been the overlord of Microsoft or IBM? But in the mobile age, they are all left behind. For the giants, a resolute and thorough reform is painful and difficult, because the profit of PC is too attractive compared with smart phones.

Otellini once mentioned his reason for rejecting Jobs in an interview: "Before the launch of the iPhone, no one knew what it could do ... At that time, Apple wanted chips within $5, which was far lower than our expected cost. In my opinion at the time, this was an irretrievable big loss-making business. "

For a giant ship like Intel, it has fully mastered the right to speak in an industry. It is not easy to change course and jump into another unfamiliar field, which also requires greater courage.

Regarding the business transformation, andy grove, the former CEO of Intel, once made an image metaphor: "Between two smoky hills, an enterprise is like a mountaineer who has to climb two hills at the same time. Successful enterprises are familiar with one mountain, but they have to run to another. On the way, the indicators are unknown, and the new mountains are looming. How long and how to get there is unknown. At this time, mountaineering teams often have fierce disputes in the valley between the two peaks. Some people want to stay in their comfortable and familiar homeland, while others want to take risks. As a result, the team fell apart and eventually died in Death Valley. "

I didn't expect one sentence to become a prophecy, which shows the fate of Intel in the mobile field.

In this long-term mobile tug-of-war, Intel was badly hurt, but it couldn't stop, missed the smartphone market and couldn't miss the AI market again. This time, it must change its course.

Therefore, internally, Intel began a series of reforms. In 20 16 years, Intel laid off12,000 people and embarked on the road of transformation again. Ke Zaiqi said that Intel should become a data company.

At present, the first half of this transformation is successful. The proportion of Intel's traditional PC business has dropped from 80% to 50% now, and the declining part has been replaced by data center business.

Externally, through the acquisition of AI chip startups Nervana and Movidius, Intel has completed the basic AI layout, but in a blue ocean AI market, the competition will only become more intense in the future.

Can Intel seize this opportunity to get on the bus smoothly? We can only wait for time to prove it.