The Achilles' heel of the giant.
By HuaShang TaoLue, Geng Kangqi
In 2000, Intel stood proudly at the pinnacle of the tech wave with a market value of $300 billion.
Over two decades later, Intel's peers have surpassed it by leaps and bounds: TSMC, which was worth $50 billion back then, now exceeds $2.5 trillion; NVIDIA, which was worth $50 billion, now exceeds $2 trillion; and even AMD, which Intel couldn't hold a candle to back then, now has a market value of over $330 billion...
Meanwhile, Intel, which once had a market value of over $300 billion, now falls short of $190 billion.
[The Most Tragic Dominator]
On March 5th, Intel made a high-profile announcement that it had received the world's first new-generation high-NA EUV lithography machine from ASML.
Priced at over $300 million, this machine is essential for mass-producing chips with advanced processes below 2nm. With this hefty investment, Intel has revealed its ambition to catch up with the global "king of foundries," TSMC.
However, as the "bottom" player that has fallen to tenth place among global wafer fabs, Intel's quest to regain its former glory seems like an "impossible mission."
In 1971, just three years after its founding, Intel developed the first commercial processor, the Intel 4004, heralding the dawn of the world's computer and internet revolution.
Since then, from memory to personal computers and server CPUs, from chip design to manufacturing, Intel has long been the absolute hegemon of the American and even global semiconductor industry.
At the end of the last century, its alliance with Microsoft formed the "Wintel" alliance, creating a wave of globalization in the PC era. In 2000, Intel's market value surpassed $300 billion, overshadowing today's NVIDIA.
But after reaching its peak, it has been virtually beaten down and has been in an unstoppable decline for nearly a decade.
In 2017, a bombshell hit the industry with the news that Samsung had surpassed Intel in performance, becoming the new semiconductor king.
Samsung, known for its focus on lower profit margins and volatile pricing in the memory chip arena, had long been overshadowed by Intel, which held high market share and profits in critical areas like personal computers and servers.
Facing this unexpected turn of events, Samsung cautiously declined to comment, while Intel, on the other hand, boldly stated, "We are very satisfied with our strategy and performance."
However, the reality was harsh for Intel: their planned 10nm chip rollout was delayed, a fatal blow in the technology-driven semiconductor industry. Meanwhile, its two biggest rivals in the CPU and foundry markets, AMD and TSMC, were making significant strides forward.
With technological setbacks and rising competitors, Intel found itself in a precarious position.
From 2018 to 2019, Intel saw record revenue and profits, but its growth lagged far behind its competitors, and its market share continued to erode.
In July 2020, already trailing behind its competitors, Intel plunged further into crisis.
The disaster came from then-CEO Bob Swan's helpless statement during a conference call: the company's future chip manufacturing plants might never catch up in progress, and they might even have to consider using contractors to manufacture 7nm chips.
Despite his attempts to remain composed, every analyst present could sense a hint of hesitation in his speech. After all, this could be Intel's most radical "bad news" in its 52-year history—a public acknowledgment of falling from the peak of high-end chip manufacturing.
Some American media even exclaimed that Intel's "stunning failure" signaled the end of the American chip era. Overnight, Intel's market value evaporated by nearly 300 billion CNY.
In February 2021, in a desperate move, Intel fired CEO Bob Swan, who had only been in office for two years. Stepping up to the plate was Pat Gelsinger, Intel's eighth CEO.
In 1985, one of Intel's founders, Andy Grove, had approached Gelsinger, then a doctoral candidate, saying, "You can fly a simulator here, or stay at Intel and fly the real plane."
More than 30 years later, Gelsinger became the "captain" of Intel, tasked with saving this once historic and pioneering giant from being completely left behind by the times.
With great ambition, he declared:
Intel is back!
As a veteran with nearly three decades at Intel, Bob Kinsley knows that "over a decade of mistakes cannot be resolved overnight."
However, he is still desperately trying to remedy the situation by announcing the IDM 2.0 strategy, which involves separating the design and manufacturing departments, adopting a more flexible wafer outsourcing model, aggressively spending on expansion and technological innovation, and reclaiming lost market share, even aiming to regain its former dominance.
To achieve this, Intel has proactively divested its non-core businesses, focusing on design and manufacturing. In March 2021, the company announced a $20 billion investment to build two new chip factories in Arizona, USA; a year later, it invested over $20 billion to build two more factories in Ohio.
In July 2021, Intel unveiled its "Four Years Five Process Nodes" plan, aiming to advance through five process nodes—Intel 7, Intel 4, Intel 3, Intel 20A, and Intel 18A (1.8nm)—within four years, aiming to regain process leadership by 2025.
During a conference call on Wall Street, Kinsley bluntly stated, "Great companies can recover from difficult and challenging periods and emerge stronger, better, and more capable than ever."
However, industry analysts poured cold water on the plan, stating, "The new roadmap is very ambitious, but it remains to be seen how it will be realized." Intel's continuous dismal performance serves as strong evidence for their pessimism.
In July 2022, Intel's Q2 financial report showed a 22% drop in total revenue to $15.321 billion, marking the first net loss of $454 million in a decade, making it the "most dismal performance in history."
In the first quarter of 2023, Intel suffered a net loss of $2.76 billion, marking its largest loss in history.
On January 26, 2024, Intel's financial report revealed a revenue of $54.2 billion for 2023, a 14% year-on-year decrease; a net profit of $1.7 billion, a staggering 79% drop year-on-year, leading to a 10% plunge in stock price.
Now, even Kinsley, who is eager to turn the tide, is forced to face reality: "There is still a long way to go to regain our former strength."
After all, Intel, which had a market capitalization of $300 billion in 2000, now has a market value of less than $200 billion. Meanwhile, Taiwan Semiconductor Manufacturing Company (TSMC) and NVIDIA, which had market values of only around $50 billion and $5 billion, respectively, at the time, now have market capitalizations of $7.591 trillion and $2.19 trillion, respectively.
Even AMD, which once couldn't even compete with Intel, now has a market value of over $330 billion, surpassing Intel.
【Retreating Step by Step】
In the 1960s, one of Intel's founders, Gordon Moore, proposed the famous Moore's Law: when the price is constant, the number of transistors on an integrated circuit will double every 18 to 24 months, and performance will also double.
Since then, Intel has been the most faithful defender of this law: firmly believing that as long as this "curse" is strictly followed, it will be invincible.
Even during industry downturns, Intel has always maintained a research and development investment of over 10%, chasing after Moore's Law. This has enabled Intel to continuously surpass its products and markets, leading to its ascent to the world's largest semiconductor company throne in 1992.
From 2005 to 2007, Intel also introduced the unique Tick-Tock production model, which operates in two-year cycles: "tick years" focusing on chip manufacturing, updating chip processes, and improving technology; "tock years" focusing on chip design and architecture innovation.
Under this framework, as long as chip performance and manufacturing processes continue to improve steadily, Intel won't squander the performance of the previous generation processors due to overly fast development pace. The substantial investments in wafer fabs can also be effectively distributed among the vast production capacity demands.
Consequently, Intel can gain more market share and financial advantage over its competitors at a lower cost, further fostering sustained growth in high-performance chip development and production capacity expansion.
As a result, Intel has secured a golden decade without wasting any opportunities or costs, steadily leading the way from 45nm, 32nm to 22nm, and 14nm development. Competitors like Samsung, TSMC, AMD, and others can only follow suit.
This tremendous success has solidified Intel as one of the few giants qualified to adhere to the IDM (Integrated Device Manufacturing) model: controlling all core aspects of semiconductor production, including design, manufacturing, packaging, testing, and sales.
Initially, global semiconductor giants like Intel, IBM, and Fujitsu all followed the "design and manufacture in-house" IDM model.
However, with the escalating costs of advanced process technology development and production, the larger firms gradually ceased investments in wafer fabs as they lost market share. The number of IDM global players swiftly dwindled from 30 during the 130nm era to fewer than 5.
Yet, the advantages of the IDM model are evident: each aspect of chip production can be properly coordinated, internal communication can reduce transaction costs, significantly shorten product time-to-market, and ultimately achieve higher profit margins.
Intel thus capitalized on this model to dominate the market, boasting a profit margin of over 60%.
However, just as Intel was enjoying its dominance, a force capable of overturning it was quietly growing.
In 1987, Andy Grove, a close friend of Morris Chang, who had just become CEO of Intel, changed the fate of TSMC, founded the same year, with one sentence: "Perhaps Intel could use you guys."
Subsequently, TSMC, which took over some of Intel's low-end chip production capacity, tasted the sweetness of foundry services. Unsatisfied with being left with only scraps from the IDM giants, Morris Chang made a decision that would profoundly impact the global semiconductor industry: to specialize in wafer foundry services.
Initially, TSMC sought to cling to Intel's coattails again but was refused financing requests. Even Gordon Moore personally advised Morris Chang, "You've had many good ideas, but this one doesn't seem great."
However, both Moore and Intel were wrong this time.
After 1991, with the arrival of the semiconductor boom and the rise of the computer industry, countless newcomers flooded into the low-barrier chip design sector. TSMC became the biggest supporter for these startups that lacked the funds and manpower to build their own wafer fabs.
By 1994, TSMC, successfully listed, had 137 clients and a market value of up to NT$120 billion.
This marked a watershed moment in semiconductor history: Intel's steadfast IDM model was continuously defeated by TSMC's specialized wafer foundry model. Many competitors who were ruthlessly crushed by Intel under the previous model were able to catch their breath and even quietly rise up.
In 1995, Huang Renxun, who had been in business for only two years, desperately sought help from TSMC as he couldn't afford to build his own wafer fab. Soon, Morris Chang called back and agreed to provide foundry services. Just as Intel changed its own destiny years ago, TSMC changed the fate of NVIDIA.
Leveraging TSMC's professional and reliable foundry model, the fledgling NVIDIA quickly captured a niche market. After proposing the GPU concept, Huang Renxun emphasized, "There is no Plan B; everything is pinned on TSMC."
Indeed, with TSMC's full support, NVIDIA swiftly became the global leader in graphics card chips. Huang Renxun later exclaimed with gratitude, "If I had built my own wafer fab back then, I might have only become a CEO with tens of millions of dollars in revenue."
Similar stories also unfolded at Qualcomm and AMD.
In 1996, as mobile phones began to proliferate globally, Qualcomm, grappling with significant research and development costs in wireless communication technology, turned to TSMC for manufacturing. This move alleviated its financial pressure, allowing Qualcomm to focus wholeheartedly on CDMA technology and emerge as an industry giant.
AMD's founder, Jerry Sanders, initially staunchly believed that "owning a fab is what makes you a real man." Despite Morris Chang's assertion that "my production costs are half of Intel's, with double the quality," Sanders remained unconvinced about outsourcing.
However, under the old model, AMD, which lagged far behind, was no match for Intel. With Intel's Tick-Tock model gaining momentum, AMD, trailing both in product design and advanced manufacturing processes, struggled to keep up and was pushed to the brink of bankruptcy.
Consequently, in 2009, AMD divested its fabs and abandoned the IDM model, opting for a path focused solely on design. Then, in 2014, AMD welcomed Lisa Su as its CEO, who further bet big on design, concentrating on developing the all-new Zen architecture. Leveraging TSMC's continuously advancing process technology, AMD swiftly caught up to Intel.
As a result, Intel found itself in a precarious position, confronting the dual challenge of competing with the world's top chip design and manufacturing companies. Step by step, it was pushed towards the abyss by its competitors.
On the manufacturing front, in 2014, Intel announced yet another delay in the mass production of its 14nm process, signaling the beginning of the end for its Tick-Tock strategy. Meanwhile, TSMC initiated the "Night Hawk Project," mobilizing nearly 400 researchers in round-the-clock shifts to accelerate the development of its 10nm process.
By 2017, Intel's 10nm plans faced significant setbacks, and its reluctantly released 14nm+ process was derided as "toothpaste squeezing." The Tick-Tock model, which had been running for a decade, was declared bankrupt. In contrast, TSMC achieved mass production of its 7nm process in 2018.
On the design front, AMD's Ryzen series CPUs, offering faster speeds at less than half the price of Intel's, disrupted the industry.
After TSMC achieved mass production with its 7nm process in 2018, AMD decisively shifted all its production capacity to TSMC. Meanwhile, Intel found itself further trapped in the vicious cycle of the IDM model: the inability to adopt more advanced processes in its own fabs hindered design innovation, leading to market and efficiency lag and dragging down investments, thus preventing the realization of advanced processes.
In July 2020, Intel announced another delay in its 7nm process, while TSMC's 5nm chips were already in production, powering devices like the iPhone 12 and Huawei Mate 40, with 3nm chips on the horizon.
At this point, TSMC's manufacturing prowess had become a nightmare for Intel, and AMD, riding on TSMC's shoulders, continued its spectacular comeback. In February 2022, AMD surpassed Intel in market value for the first time and now commands over 20% of the global PC and server market, rapidly eroding Intel's future prospects.
In this sense, Intel's failure was a failure of its model, its attempt to single-handedly compete with global rivals in both design and manufacturing, ultimately leading to doing everything yet excelling at nothing.
However, beyond the model, Intel also had larger factors contributing to its downfall.
[Loss of Future]
Grove, who led Intel to its peak, once pointed out in "Only the Paranoid Survive" that being able to identify the direction of the wind and avoid shipwreck is crucial for a company's future.
Yet, the subsequent Intel repeatedly missed the "direction of the wind" for the future.
At the 2006 Macworld Conference, a classic scene unfolded in the annals of tech history:
Wearing semiconductor cleanroom attire, Intel CEO Paul Otellini handed a large silicon wafer to Steve Jobs, exclaiming enthusiastically, "Intel is ready."
Over the past decade, Intel has dominated the vast majority of the world's personal computer CPU market, while Apple remained the last unconquered territory. Otellini, who was praised by the "Silicon Valley Idol" Steve Jobs, won a huge victory for Intel.
Since then, the two became close allies, but Otellini won the beginning only to lose in the end.
When the first generation of iPhone products was in the making, Jobs personally approached Otellini, hoping that Intel would manufacture chips for Apple's smartphones. However, this time Otellini made a historic mistake.
He carefully organized internal experts to analyze the possibility of manufacturing chips for the iPhone. The experts unanimously believed that Jobs' order price was too low and the quantity was limited, making it an unprofitable deal.
Otellini listened to the advice of these professionals and said no to Jobs.
The experts' considerations were also correct. At that time, Intel already had an absolute share in the X86 chip market for personal computers and servers, with profits rolling in like a money-printing machine.
However, the emerging field of mobile chips, dominated by devices such as smartphones and tablets, favored the lower-power ARM architecture products. Intel found it difficult to provide high value-added products, and investing heavily might not be worth it.
In fact, Intel even proactively sold its XScale business, which produced ARM architecture products, and developed a low-power x86 chip called Atom to target the ARM products entering the laptop market, further defending its dominance in the computer field.
In short, Intel chose to concentrate its efforts on guarding the PC market, lacking interest in the mobile market.
A chance to win the next wave of industry competition was thus personally buried by Intel.
In the end, the sales of Apple's first-generation iPhone exceeded the initial estimates by more than 100 times. Apple's revenue from smartphones surpassed Intel's in just a few years.
Choosing the wrong future not only caused Intel to miss out on the biggest ticket of the mobile Internet era but also led it to bury itself in the continuing decline of the personal computer market. Otellini deeply regretted rejecting Jobs' offer, calling it:
"The most regrettable thing in my career."
But this might not be Intel's most painful lesson.
In 2006, NVIDIA, relying on GPU trends, had already begun to show signs of changing the future.
That year, with AMD's strong acquisition of the GPU "runner-up" ATi, NVIDIA immediately launched the Tesla GPU with CUDA architecture, whose performance was as powerful as a "nuclear weapon" in the computing field.
The high-profile NVIDIA made Intel wary. To thwart NVIDIA's progress, Intel not only terminated their integrated graphics cooperation but also introduced the General Purpose GPU (GPGPU) and announced a major decision: to launch the Larrabee project.
This project, said to have cost billions of dollars, was a rare charge by Intel into the peak of GPU technology. Its leader was Pat Gelsinger, the first CTO of Intel who led flagship products like the 80486 processor and is now the CEO.
However, in its impatience, when product development faced setbacks and couldn't land as scheduled, Intel, feeling that the investment was like a black hole and even if successful, there was not much foreseeable gain, firmly "cut off" the project.
Moreover, they even kicked out Gelsinger, who had staunchly supported the project.
After cutting off the Larrabee project, Intel also had a chance to counterattack: they once decided to follow AMD's lead and buy NVIDIA.
But this opportunity to change their fate, Intel missed again.
At that time, NVIDIA's market value was only around a few tens of billions of dollars. Buying NVIDIA was not a problem for Intel in terms of price, but it was said that the company's management could not reach a consensus on "what position to give to Huang Renxun," and the acquisition eventually fell through.
Today, Gelsinger, who has returned to lead Intel again, still occasionally harbors resentment about the past. "When I was kicked out of Intel, they killed a project that would have changed the landscape of artificial intelligence."
Facing NVIDIA, with a market value surpassing $2 trillion, Gelsinger still refuses to accept it, even believing that "NVIDIA's success in the AI industry is purely accidental," and Huang Renxun is "extremely lucky."
But Bryan Catanzaro, a former NVIDIA executive who worked on the Larrabee project, disagrees with Gelsinger's statement. In his view, "NVIDIA's dominant position does not come from luck. It comes from vision and execution, which is exactly what Intel lacks."
[Reference Materials]
[1] "Revelations of Intel: Losing to TSMC, Losing to Themselves, and Losing to the Times" Jin Duan
[2] "Chip Wars" Huake Publishing House
[3] "Chip Tide" Electronic Industry Press
Welcome to follow [华商韬略] for insights into influential figures and legendary strategies.
All rights reserved, unauthorized reproduction prohibited.
Some pictures are from the internet.
If there is any infringement, please contact us for deletion.
