Author | ZeR0
Edit | Desert Shadow
One thing Intel's fifth CEO, Paul Oudelling, regretted when he retired was not saying "yes" to Apple's iPhone.
Unfortunately, there is no regret medicine in the world, the wrong decision made Intel and the mobile revolution pass by, and the rejected Apple turned around and supported the unknown Arm company, achieving the "ace chip designer" who dominated the rules of the game in the mobile Internet era.
From 2001 to 2016, in fifteen years, Apple redefined five consumer electronics tracks: the iPod revolution MP3, the iPhone revolutionizing mobile phones, the iPad revolutionizing laptops, the Apple Watch revolutionizing electronic watches, and the AirPods revolutionizing wireless headphones.
These game-changing disruptive products have built-in chips based on the Arm architecture.
Looking back, the intertwining of Apple and Arm's fates goes far beyond that. Arm from the birth to determine the authorization model, to open up a new territory, are inseparable from its "father" Apple's assistance; behind the growth of Apple's chip empire, Arm is also the protagonist who can never be avoided.
But when Apple sat in the position of the world's top chip design company, Arm's fate was in turmoil, and the road ahead was full of fog.
First, Arm was born in the hands of Apple peers
If Apple hadn't extended an olive branch to Icon Computer, whether Arm could appear on the stage of history as an independent company may have to be marked with a question mark.
In April 1976, Steve Jobs and Steve Wozniak founded Apple Computer Inc. in Cupertino, California, to focus on the development of consumer-grade personal computer (PC) products. Three years later, Hermann Hauser, an Austrian doctor of physics, and Chris Currie, a British engineer, founded Acorn Computers, a computer company in Cambridge, England.
▲ Icon Computer founders Herman Hauser and Chris Currie
Acorn means acorn in English, and there is an interesting rumor about the origin of its name, saying that the founders wanted the company to rank ahead of Apple in the telephone yellow pages. Later, for his contributions to the early days of the PC revolution in the 1980s, Icon was also known as the "Apple of Britain".
The emergence of Arm design starts from the opportunity of Icon to develop the BBC Micro computer.
In 1981, the BBC planned to launch a computer popularization program in the United Kingdom, so it widely distributed "hero posts" and wanted a standard computer to match the program. The Icon team seized the opportunity to build the BBC Micro, a miniature computer that met the needs of the BBC, in just a week or so.
▲BBC Micro
But the thorny problem was still in front of Icon, the BBC asked the computer to sell for no more than 500 pounds, but at that time, the cost of a Motorola 6800 series microprocessor was as high as 100 pounds, and it was impossible to control the price of the computer to less than 500 pounds.
IKON wanted to ask Intel to provide technical information on the 80286 chip, but was refused. Just as it was running into a wall, its engineers saw a white paper on the Berkeley RISC Program from the University of California, Berkeley, which gave the Icon team the idea of developing a chip themselves.
Instruction Set Architecture (ISA) is a standard language that bridges the communication between chips and software operating systems, and the x86 instruction set architecture that is popular in the PC processor industry is a Complex Instruction Set Operation (CISC) architecture for implementing highly complex hardware execution commands.
In contrast, Berkeley's proposed reduced instruction set (RISC) is more concise, elegant, and flexible, which can greatly reduce the complexity and cost of chip design for specific tasks. This is exactly what iCkern needs for chips.
In October 1983, the Acorn RISC Machine program for Acon's Reduced Instruction Set Architecture Processor was launched. After a year and a half of development, the first chip based on the Arm architecture, ARM1, was finally born, which is also the world's first commercial chip based on a simplified instruction set.
▲Acorn RISC Machine
At 15:00 UK time on April 26, 1985, the first batch of computers with ARM1 chips successfully started, and the boot screen showed: "Hello world, I am Arm".
Arm chip in the BBC Micro computer
This is the starting point of Arm's story, and today, 37 years later, it will dominate far beyond The United Kingdom and spread across the world.
Second, Apple and Icon "have children" to build a core for Newton's computer
At the beginning of the Arm architecture, Intel x86 architecture has formed a "WinTel" alliance with Microsoft's Windows operating system, and has gradually become the gold standard in the global PC industry.
Compared to the x86 architecture, the Arm architecture has lower power consumption and cost, but it is difficult to match in terms of performance. By 1985, Arm chips were doing mediocre in the PC market, Aikang and Apple were also selling poorly in the PC field, and Jobs was removed from the apple board.
Two frustrated computer companies soon intersected. In the late 1980s, when Apple was looking for a suitable mobile processor for its upcoming Newton MessagePad, apple's chip foundry partner, Introduced Icon, which has a low-power, high-performance chip.
Apple launched the First Newton Message Pad, a device equipped with an Arm chip, in 1993
The two sides hit it off, and Larry Tesler, Apple's chief scientist at the time, immediately decided to use a customized Version of arm chips in Newton's handheld computer that was adjusted to the needs of the product.
It takes overhead to make adjustments, but Icon doesn't have enough money. So in the fall of 1990, Apple, VLSI and IKON spent six weeks negotiating the joint venture, and finally decided that Apple would pay for it, IKon would pay for it, and VLSI would provide tool technology, and the three companies would jointly create a chip joint venture, and the company name was changed from "Acorn RISC Machine" to "Advanced RISC Machine", referred to as Arm.
On November 27, 1990, in a barn in Cambridge, England, Arm began its legendary journey.
Arm's initial office was in a barn
At the beginning of its establishment, Icon and Apple each owned a 43% stake in Arm. In addition to the capital, Icon also provided 12 engineers as the initial members of Arm, and VLSI contributed £25 to cover Arm's chip foundry production needs.
Apple's Newton handheld orders supported Arm's initial livelihood. But because Apple overestimated the technology available at the time, Apple's Newton pocket computer market did not perform well. This made Arm executives realize that to stay successful, you can't rely on individual products.
So Robin Saxby, then Arm CEO, decided to use the IP business model, which was to license his chip design intellectual property (IP) to more chip companies, and earn a living by collecting upfront license fees and royalty. Since then, chip design practitioners around the world can be Arm's customers.
▲Arm's business model
During this period, the division of labor in the global semiconductor industry chain began to be refined. The rise of Taiwan's wafer foundry TSMC has changed the vertical integrated production model (IDM) of the entire industrial chain of traditional chip companies, and the business model of IP licensing, chip design, foundry, packaging and other links has entered the historical stage.
In the near future, Apple's investment in Arm will help Apple come back to life and bring fruitful returns to Apple in the future.
Third, Apple sells Arm shares to save itself
In 1990, Icon was acquired by the THEN IT giant Olivetti due to its operational crisis, and Arm was completely independent.
In the new decade, the young Intel and Microsoft joined forces to dominate the PC market, ending the computer era ruled by their former big customer, the previous computer "brother" IBM, and becoming the world leader in microprocessor and PC software, respectively.
Based on Andy Grove's report at Intel (Source: Strategic Thinking)
Apple computers, which used IBM PowerPC chips and their own operating systems at the time, were unable to resist the storm brought by the WinTel Alliance. Apple's PC market share declined all the way, and by the mid-90s, it was already facing huge losses.
When Jobs returned to Apple in 1997, Apple was only nearly $400 million in profits. In order to reverse Apple's dismal operation, Jobs made two key moves over the next two years: first, pull Microsoft to invest $150 million in Apple when it was only 90 days away from bankruptcy; second, stop the production of Newton's handheld computer and start selling Apple's arm stake.
From 1998 to 2003, Apple sold its stake in Arm for $1.1 billion. The life-saving money from Arm's shares was used by Apple to advance the development of the new digital music player iPod.
The first generation of iPods, which appeared on October 23, 2001, became an epoch-making revolutionary product with its unique appearance and humanized operation. The success of the iPod series not only pulled Apple from the brink of death from a niche brand to the mainstream market, but also expanded Apple's product range from computers to consumer electronics. Successive iPod products have used Arm chips.
At the same time that Apple is struggling to get out of its own predicament, the emerging mobile phone market has risen, and a door to the future mobile era is slowly opening to Arm.
The core appeal of PCs to chips is high performance and ability to run complex applications, but when the market began to turn to smart phones and tablets, the rules of the game changed, and low power consumption, power saving, and strong endurance became the new demands of consumers. This appeal is precisely the shortcomings of the x86 architecture and the strengths of the Arm architecture.
At the same time, Arm's IP licensing model has formed a "win-win" symbiotic relationship with its partners. Arm-licensed chip companies can focus on the design and sales of overall chips and dilute research and development costs. Arm, as the exporter of core technology, stands at the top of the entire chip design industry value chain and does not need to bear the risk of chip design failure.
With the advantages of low-power technology and a pioneering IP licensing business model, Arm's shipments in the field of mobile chips began to explode and quickly dominated the global mobile phone market.
▲Arm processor shipments from 2000 to 2014 (Source: recode)
In 2005, 98 percent of the world's mobile phones were equipped with at least one Arm chip. In the next 15 years, Arm became a chip standard for the unified mobile terminal consumer electronics market. Pc chip overlord Intel is in the trap of innovators.
Fourth, the era of unified mobile Internet: Arm everywhere
On June 6, 2005, Intel's fifth CEO, Ou Dening, put on Intel's iconic Bunny Suit tooling and appeared at Apple's WWDC conference, announcing with Jobs that Apple Computers Macs will be replaced by Intel chips.
Jobs (left) and Oudenin (right) announced that the Intel Mac will be equipped with Intel chips
This is another triumphant moment for Intel in the PC space. Immersed in joy, Oudenin ignored an epoch-making change: the PC era reached a bottleneck, the smartphone era rushed to take over the baton, and the new big customer Apple is the key engine to launch the innovation revolution.
The advent of a revolutionary product that changed and redefined the phone, the first iPhone, was in the works, and would quickly render industry giants Nokia and BlackBerry insignificant.
It is lamentable that in the 80s, when Japanese companies were forced to a desperate situation in the field of memory chips, Intel seized the opportunity of the rise of PCs to transform microprocessor suppliers and ascend to the throne of global semiconductor hegemony. But when Apple approached Intel in the hope that it would provide mobile chips for the iPhone, Intel failed to have enough awe of the emerging field that opened its doors to itself.
Ou Dening believes that the future of the iPhone cannot be determined, Apple's price is lower than expected, and this thinly profitable transaction is not cost-effective for Intel, which can make a lot of money on PC chips. Years later, Mr. Oudenin recalled the decision with great regret: "In hindsight, we look back and see that the output has been many times higher than the input, and the iPhone's performance has obviously exceeded everyone's expectations." ”
Intel also had an Arm-based XScale chip business. Speaking of which, XScale is also related to Apple, whose predecessor was the StrongARM architecture that Intel acquired from DEC in 1997, and Apple used StrongARM chips in its Newtonian mobile devices.
XScale was once the dominant player in the field of handheld computer chips, and also developed an Arm-based mobile phone chip, which was called the second largest smartphone chip in the industry after Texas Instruments chips at the time. However, nearly a decade after advancing the XScale business, Intel sold it to the American chip company Marvell for about $600 million in 2006. In the future, Intel will try to cut into the mobile field, and it will be too late.
In early June 2007, Apple released the revolutionary iPhone, and the mobile Internet reached dawn. Five months later, Google launched the first version of the open source Android operating system, and Qualcomm launched the first Snapdragon processor, the Snapdragon S1, using the Armv6 architecture.
After four years of market fighting, Google Android system has become the world's largest smartphone operating system, Qualcomm Snapdragon processor is the chip supplier of the vast majority of Android mobile phones, Google Android and Qualcomm have been given the new name of the WinTel Alliance for the benchmark PC market - the mobile phone market Quadrioid.
▲Changes in the market share of smartphone operating systems from 2009 to 2018
In the four years of this smartphone market reshuffle, Apple has also begun to prepare for its own chip team. Since then, no matter how the global mobile chip market changes, Apple, Qualcomm, Huawei, MediaTek, Tsinghua Unigroup Zhanrui... The core chip products of several major head players are inseparable from the authorization of arm architecture.
Fifth, son Zhengyi's heart to buy arm seeds, was planted by Jobs
In Japan, nearly 10,000 kilometers from Britain, An Internet service provider, SoftBank Group, began plotting to buy Arm.
In 2006, SoftBank Group spent $15.5 billion to acquire the Japanese subsidiary of global mobile communications giant Vodafone in the United Kingdom. Shortly before the acquisition, SoftBank founder Masayoshi Son rushed to Apple's headquarters with a sketch of fusing the iPod with a Japanese mobile phone to meet with his old friend Jobs.
▲ Son Zhengyi (left) and Jobs (right) group photo
Jobs glanced at the sketch and, under the premise of implementing the principle of extreme confidentiality for the new product, gave Son a hint: "The time has come to make the strongest mobile device, and I fully agree with this." ”
This meeting laid the groundwork for SoftBank's exclusive sale of iPhones in Japan, and also prompted Son Zhengyi to start thinking: What kind of CPU is used for the "brain" of Jobs's strongest mobile device?
According to Son, Jobs, a person who has the ultimate pursuit of beauty, "would never choose the ugly design of the battery protruding like a toad." After stripping away the cocoon layer by layer, he concluded: "Arm will master the platform in the era of mobile Internet." ”
Optimistic about Arm's future Son Zhengyi, let the SoftBank related team study the acquisition of Arm. But SoftBank, busy with the rebuilding of Vodafone's Japanese subsidiary, shelved the idea of buying Arm, a shelving that came a decade later. It wasn't until July 2016 that SoftBank announced its intention to buy Arm for £24.3 billion.
▲ As of February 2020, Arm's chip shipments exceeded 160 billion
In the decade between SoftBank's initial plans and the formal acquisition of Arm, Apple has built a chip empire covering the entire product line based on the Arm architecture.
Apple's first self-developed Arm chip
From time to time, history enters reincarnation. Just as hp five companies in the 70s refused to produce computers designed by Wozniak and Jobs, forcing the two to set up Apple Computer Company to build their own computers, the disagreement between Apple and third-party chip companies in chip design made Jobs begin to consider forming Apple's own chip research and development team.
Apple's first three generations of iPhones were all equipped with arm-based custom system-on-chips (SoCs). Previously, Samsung had produced simplified Arm chips for Apple's tens of millions of iPod players, and Apple also outsourced the chip business of the original iPhone to Samsung, but when negotiating with Samsung, Apple's weakness was exposed - there was no chip expert of its own, and there was insufficient understanding of chip design, and as a result, the APL0098 chip produced by Samsung for the original iPhone had average performance.
Jobs realized that to maintain Apple's lead, he had to have his own chip design team. On the one hand, Apple has reached a licensing agreement with Arm and Imagination, a British GPU IP supplier, so that Apple can develop and optimize mobile chips with Samsung, and on the other hand, it actively absorbs chip talents and continuously strengthens chip research and development strength.
In 2010, Apple's revolutionary product tablet iPad came out, and surprisingly, the iPad did not use Intel's chips like Apple's Mac, Apple TV set-top boxes and other product lines, but was equipped with Apple's self-developed A4 chip based on Arm architecture. In addition to the iPad, the chip was also used in the iPhone 4 as well as the redesigned iOS version of the second-generation Apple TV set-top box.
At the beginning of the iPad project, Jobs originally planned to use the Intel Atom chip in the iPad, but Tony Fadell, the hero of the development of the iPod and an Apple engineer who loves cyberpunk styling, vigorously opposed this idea and supported the Arm chip, which is good at low power consumption.
During one meeting, when Jobs insisted on trusting Intel to make the best mobile device chips, Fadle yelled, "Wrong, wrong, wrong!" He even slapped his name tag on the conference table and threatened to resign.
Eventually, Jobs followed Fadell's advice. Of course, another reason for this decision is that it is difficult to reach an agreement with Intel.
Jobs recalled: "For years, we reported to Intel that their graphics chips were terrible. "In the beginning, Apple and Intel made great things together, but this collaboration did not continue for two reasons: First, Jobs thought Intel was too slow, like a steamship, not very flexible; second, he did not want to teach Intel everything because he was worried that Intel might sell Apple's things to its competitors.
Apple acquired P.A. Semi, a 150-person microprocessor design company, in 2008 and licensed Arm chip technology in the same year, thus beginning the road to self-developed chips based on the Arm architecture.
In 2013, Apple launched the first truly mass-produced 64-bit Armv8 processor A7 chip, which uses the 28nm process and the new Cyclone core, which is the best in the entire mobile application processor field.
▲Apple A7 chip
Whether it is for Apple or the chip industry, this is just the beginning of a change.
On the one hand, Samsung's rapid development and the plagiarism of the iPhone incident have made its relationship with Apple worse and worse, and Apple has instead handed over a huge order for the 20nm A8 processor to TSMC. At this time, Intel has begun mass production of 14nm, the chip process is ahead of TSMC, and the olive branch thrown by Apple has boosted TSMC to step up research and development and expand production capacity.
In the future, when TSMC took the lead in achieving 7nm mass production and sat on the wafer foundry half of the country, Samsung's process research and development rhythm was always one beat slower than TSMC, and Intel's 10nm products had not yet landed in the market on a large scale.
On the other hand, Apple's quietly accumulated chip research and development strength has gradually emerged, and it has begun to quickly leave a group of mobile application processor designers far behind.
Seventh, take advantage of Arm, the Apple chip empire that rises on the ground
In 2014, Apple further optimized the 64-bit Cyclone architecture on the A8 and A8X processors, using a 20nm process process.
At this time, Samsung and Qualcomm have not yet produced 64-bit Arm chips suitable for mobile phones, and Intel has not been able to show strong attraction in the mobile chip market. The following year, Intel stopped announcing the results of its mobile division in its earnings report, and Nvidia simply abandoned the mobile device market.
And Apple chips are still striding forward, smart phones, tablets, smart watches, wireless headphones, laptops and desktop hosts... Based on the Arm architecture, Apple's self-developed universal chip architecture has successively reached each of its product lines.
▲ In 2014, Apple's first generation of wearable smart watch Apple Watch was equipped with Apple's self-developed S1 chip
After 2016, Apple's first-generation wireless headset AirPods equipped with its self-developed W1 chip can realize wireless connection management and quick pairing and connection with iPhones and iPads. This year, Apple also began to use the T series of security coprocessors in the new MacBook Pro computers to control the new Touch Bar touch bar, Touch ID secure encryption and other tasks.
In September 2017, Apple launched the A11 Bionic, the first self-developed AI bionic chip for the iPhone, which for the first time was equipped with Apple's self-developed triple-core GPU and Neural Engine to accelerate AI tasks. To this day, Apple's A-series chips have always been the everest in the field of mobile application processors, making it difficult for other opponents to match.
▲2010-2017 Apple's self-developed chips equipped with various terminal devices
Apple's core-making territory is still expanding. In March 2019, Apple released a new AirPods after three years using a new H1 chip, compared with the previous W1 chip, H1 strengthens the wireless connection performance and endurance. In June 2020, Apple announced that it will introduce self-developed Arm architecture chips in Mac products, and it is expected that the transition period from Intel chips to Apple's self-developed chips will be two years.
Standing at this point in time, Johny Srouji, apple's senior vice president of hardware technology, announced the phased results: In just a decade, Apple has delivered 2 billion SoCs and billions of other chips.
Today, Apple's "two-year core change" plan is only one step away. After Apple announced in March 2022 that its host Mac Studio uses the M1 Ultra, the flagship chip of the M1 series, only waiting for the Mac Pro to change the core, all of Apple's Mac product lines will fully apply the self-developed chip.
▲Apple's self-developed M1 series computer chips, from left to right: M1, M1 Pro, M1 Max, M1 Ultra
However, when Apple steadily released new chip moves, Arm's fate became like fluttering in the wind.
Eighth, the dream shattered the Internet of Things, Arm returned to the listing road
SoftBank founder Masayoshi Son's IoT dream dragged down Arm's search for a growth curve outside the mobile phone market.
At that time, Son Zhengyi was full of enthusiasm and believed that artificial intelligence and the Internet of Things were the future trends. According to the predictions of relevant agencies at the time, by 2020, the number of Installed IoT devices will reach 26 billion. If Arm can conquer the vast majority of Internet of Things devices, the benefits can be imagined.
This is also why SoftBank proposed an all-cash purchase price of more than 40% premium to the highest value in the history of Arm stock when acquiring Arm, and even sold Ali's shares to raise the transaction money.
In 2016, when SoftBank acquired Arm, Arm's most stable market, the global smartphone market, the shipment growth rate has fallen to single digits, and in 2017-2020, it has declined year-on-year, and it will not return to single-digit growth until 2021.
In order to create new revenue streams in areas such as the Internet of Things and artificial intelligence, Arm began to expand its team with the support of SoftBank.
In fiscal years 2018-2020, arm teams increased by 1,034, 101 and 751 respectively, and the number of employees increased by 7.9% in fiscal 2021, and operating costs continued to rise with the expansion of the team size.
But the resources tilted to the Internet of Things have not brought Arm a huge profit. The fragmented IoT device market is not developing faster than expected, and a number of executives who have been in charge of Arm's IoT business revealed after leaving that Arm's IoT business is progressing slowly.
SoftBank's earnings data shows that Arm's costs grew from $716 million in 2015 to $1.6 billion in 2019, but its revenue growth slowed, with annual revenue growth of less than 5% in fiscal years 2017-2020. In fiscal years 2015-2019, Arm's net profit was $843 million, $827 million, $436 million, $279 million and $276 million, respectively, which has turned into a loss by fiscal 2020.
▲Changes in Arm's revenue and adjusted EBITDA from 2015 to 2020 (Source: Financial Times)
In July 2020, Arm announced the spin-off of its two IoT service businesses, IoT Platform and Tree Data, and transferred them to SoftBank Group, with itself focusing more on its core IP business. Within a week, foreign media broke the news that the already heavily indebted SoftBank Group was studying various options, including selling Arm in whole or in part, or re-listing Arm.
This was followed by the NVIDIA acquisition of Arm that sensationalized the global tech circle. In September 2020, Nvidia announced that it would acquire Arm from SoftBank for about $40 billion, a deal scheduled to close within 18 months.
Since the announcement of this transaction, many of Arm's old customers have publicly opposed, although Nvidia has repeatedly promised not to change Arm's openness and neutrality, nor will it move Arm's base camp from the United Kingdom to the United States, but it is a matter of business security, many Arm customers do not buy it.
Ultimately, the huge cross-border M&A transaction came to an end in February 2022, with Nvidia and SoftBank announcing that the deal was terminated due to major regulatory challenges that hindered the completion of the deal. Soon after, Intel CEO Henry Kissinger said that if there is a consortium to include Arm, Intel is interested in participating.
In the past year of advancing its M&A deal with NVIDIA, Arm has corrected its focus from the little-impact IoT to data centers, networking equipment and automotive with more large customers. Thanks to the growth in shipments of 5G smartphones, 5G base station network equipment and servers, its global business began to regain growth momentum. However, compared with the x86 architecture, Arm's market share in the server field is still "small and big".
▲Changes in Arm net sales distribution from FY2017 to FY2021 (Source: Statista)
To ensure a reasonable balance between market opportunities and cost constraints, Arm announced in March 2022 that it would cut 12% to 15% of its workforce globally, involving around 1,000 employees, most of whom work in the UK and the US.
Arm currently employs around 6,400 people worldwide, of which 3,500 are in the UK. According to foreign media reports, rene Haas, the new Arm CEO, promised in a memorandum to employees that most of the layoffs would not affect engineers.
Now, Arm is preparing to restart its listing by March 2023, and this cornerstone of the mobile computing space will once again be put in front of capital for sale.
IX. Conclusion: The resonance of the fate of two business legends
At that time, the light train drove through the golden years of the mobile Internet, and the two technical leaders of Apple and Arm, under the symbiotic relationship of mutual support, the fate of the entanglement and resonance, achieved each other.
Apple's demand for innovation has brought the first pot of gold to arm's rise, and with its industry leader influence to help Arm expand its ecological territory; Arm's technological evolution has accelerated the construction of Apple's chip empire.
As the economist Joseph Schumpeter put it, the economic character of the modern business revolution has always been accompanied by two main characteristics: "never-ending winds" and "creative destruction." The wave of technology is always new, and Apple and Arm, which challenged the old order and led the new revolution in technology, have now changed their identities into defenders.
While Arm's book of fortunes is full of unknowns and Apple is under pressure to consolidate its position as an innovation leader, there is no doubt that these two companies at the height of the mobile era are surging with a desire for change, constantly pushing boundaries and trying to add new value to their past glories.