This year's Taipei International Computer Show (Computex) gathered many industry big names. The CEOs of the world's five largest chip giants, Nvidia, AMD, Qualcomm, Arm, and Intel, all came to the exhibition and gave keynote speeches.

The theme of this year's exhibition is undoubtedly AI PC. Major chip giants have demonstrated the new AI experience brought to the PC industry by their new products, hoping to boost the PC industry, which is in a trough, to regain growth. According to Gartner data, global PC shipments fell 14.8% to 228 million units last year, falling below the 250 million mark for the first time since 2006.

Nvidia Moore's Law

As the co-founder and CEO of Nvidia, the leading AI stock that has been enjoying great success recently, Huang Renxun, who returned to his hometown, was the biggest focus of the exhibition. In his keynote speech, he looked forward to Nvidia's product technology roadmap for the next three years. Following the recent release of the Blackwell chip, he introduced the next-generation chip Robin.

Although Nvidia did not release too many details about Robin, Huang Renxun looked forward to the arrival of the new industrial revolution of AI with great ambition. He mentioned that with the global computing energy consumption and cost soaring, Nvidia innovatively combined CPU and GPU to bring solutions for accelerated computing, thus achieving cost-effective computing acceleration.

This technology leader who still wears leather jackets in summer also proudly challenged Moore's Law, the "eternal truth" of the semiconductor industry. He expects that Nvidia's innovative architecture AI chip will achieve a 1,000-fold increase in computing power, floating-point computing, and AI computing power within eight years, significantly reducing computing costs, thereby breaking Moore's Law of traditional processor architecture.

With a market share of more than 80% in the field of AI processors, Nvidia has been the biggest winner in the AI boom in the past two years, with its stock price rising by more than 550%. 

Just after Huang Renxun's speech in Taipei, Nvidia's market value once exceeded the 3 trillion US dollar mark this week, standing on the same level as Microsoft and Apple.

With the advent of the AI era, Nvidia has rapidly risen to become the global semiconductor innovation leader, while the former chip giant Intel seems a little lonely and embarrassed. In the past few years, Intel has been trying to get out of the predicament of industry decline through strategic transformation and technological innovation.

In the past sixty years, Intel has been a landmark company in the semiconductor industry in Silicon Valley. In the 1980s, Intel and Microsoft joined hands to open the PC era and became an industry giant in one fell swoop. In contrast, AMD has always lived in the shadow of Intel and was even on the verge of trouble. In the first 25 years of the company's establishment, Nvidia basically relied on gaming graphics cards to make "small money".

The chip giant is unwilling to accept its decline

Times have changed. Now is the AI era. Nvidia is soaring, AMD is also turning over. Intel now has the lowest market value among the five major chip giants, and is even worse than Arm, which only designs chips. Nvidia's current market value is equivalent to more than 20 Intels, and its market value can even soar by $140 billion in one trading day, which is equivalent to easily increasing the value of one Intel during the trading session. If there is no antitrust shadow, Nvidia can easily buy all chip companies.

Several other chip giants, even if they do not have the crazy growth of Nvidia, have also achieved good performance and stock prices in the AI era. So far this year, AMD's stock price has risen by 20%, with a market value of nearly $270 billion; Qualcomm's stock price has risen by 51%, with a market value of nearly $237 billion; Arm's stock price has risen by 98%, with a market value of more than $143 billion, and Masayoshi Son's SoftBank has made a comeback.

In sharp contrast, Intel, the former Silicon Valley chip overlord, has seen its stock price fall by 36% this year, and its market value is now even less than $130 billion, making it the worst performing component of the Philadelphia Semiconductor Index. AMD, which was once suppressed and could not lift its head, has now made a big comeback, with a market value equivalent to two Intels. This is undoubtedly a great humiliation for Intel.

Intel's revenue fell sharply by 14% last year, from US$63.1 billion to US$54.2 billion, but it is second only to Nvidia among these major chip giants, and more than twice that of AMD. However, Intel is also the cheapest among the five giants, with a price-to-earnings ratio of only 31.7 times. In contrast, Nvidia's price-to-earnings ratio is 71.4 times, and Arm is as high as 471 times.

The stock price is actually investors' expectations of the future growth space of a listed company. Behind the crazy surge in Nvidia's stock price is their absolute dominant position in the AI processor industry, while the fundamental reason for Intel's sluggish stock price is that they are constantly hit in the PC field and have lost their competitiveness in the new AI era. In the last fiscal quarter, Intel's data center business revenue fell by 10% to only US$4 billion, which was lower than market expectations; while Nvidia's soared 427% to US$22.6 billion, which was higher than market expectations.

Kissinger defends Moore's Law

While Huang Renxun stole the limelight at the Taipei Computer Show, Intel CEO Pat Gelsinger was slightly embarrassed. On the one hand, he needs to maintain Intel's position in the chip industry, show that his chips are still competitive, and deal with the strong impact of competitors. On the other hand, he has to admit that Intel's current chip process has technical defects, and even has to let competitors manufacture the next generation of flagship chips.

Moore's Law is a prediction of semiconductor performance multiple growth proposed by Intel's legendary co-founder and CEO Gordon Moore in 1965 (originally doubling every two years, later revised to 18 months). Although more than half a century has passed, this law is still the golden rule of the semiconductor industry and represents Intel's glorious history in Silicon Valley.

After Huang Renxun challenged Moore's Law, as Intel CEO, Gelsinger must maintain the pride of this giant. He then directly responded to Huang Renxun's statement, "Unlike what Huang Renxun said, Moore's Law is still alive and well." He also emphasized that Intel will also play an important role in the popularization of AI.

Kissinger brought Intel's latest data center processor, the "Xeon 6", to the exhibition. This processor, code-named "Sierra Forrest", is based on Intel's 3 process technology and can provide efficient computing power while significantly reducing energy consumption. 

Compared with the second generation, it can provide 4.2 times the rack performance improvement and 2.6 times the performance per watt improvement.

In addition, Kissinger also unveiled Intel's new flagship PC chip processor Lunar Lake, which is Intel's main processor for the AI PC era and is scheduled to be launched this fall. According to Intel, Lunar Lake integrates LPDDR5X memory directly into the chip package, which can reduce SoC power consumption by up to 40%, thereby improving data processing efficiency and speed.

This is Intel's weapon to deal with the impact of Arm chips such as Qualcomm in the PC industry. Intel specifically emphasized that Lunar Lake's absolute performance exceeds Qualcomm's similar products, increasing AI performance by three times and overall performance by four times. Compared with the previous generation of chips, Lunar Lake CPU performance is improved by 14%, graphics performance is improved by 50%, and battery life is improved by 60%.

Intel's lost decade

Just a few years ago, it was hard to imagine that Intel would publicly compare performance with Qualcomm at the launch of its flagship chip, because the x86 architecture used to have an absolute advantage over the Arm architecture in terms of chip performance. But now, Intel has to fully cope with the huge impact of Arm chips represented by Qualcomm on the Windows PC market.

According to Canalys statistics, Intel's market share in the global notebook chip field was still as high as 78% last year, AMD took 13% of the share, and Apple, Qualcomm and MediaTek and other manufacturers had a total share of only 9%. Although PC market shipments continue to shrink, this is still regarded by Intel as its absolute territory.

As early as 2012, Microsoft launched the Surface RT, a two-in-one notebook based on the Arm architecture, and Qualcomm also launched a notebook chip designed for WIndows in 2017. 

But at that time, these Arm architecture notebooks could not be compared with Intel chip products in terms of performance or compatibility, and they did not have the conditions for popularization.

However, with the substantial improvement of chip technology in the past few years, especially the rapid progress of process technology, the performance of Arm architecture chips is far from the weak level of the past, while Intel has fallen into the bottleneck of its own process technology, and has not been able to break through the bottleneck of 10nm for several years. Intel, which was the first to break through the 14nm node in 2014, fell into a lost decade, watching TSMC and Samsung far surpass itself.

Apple has switched to its own M series processors based on Arm architecture since 2020. With its internal customized design and the inherent advantages of 5nm, its performance directly overwhelmed Intel's chips that were still spinning in the 10nm process, which directly led to Intel gradually losing all Mac chip orders.

Last year, Qualcomm upgraded and released the Snapdragon X Elite, which put Intel in an embarrassing situation. For the first time, the Snapdragon X Elite has improved the performance of Arm chips to a level that overwhelms Intel Core Ultra. In addition to continuing the traditional advantages of always being online and connected, it has also added high computing power and AI performance to PC processors.

After performance is no longer a problem, compatibility has become the last obstacle to the popularization of Arm chips. Earlier this year, Qualcomm publicly guaranteed that its Arm-based notebook chips can seamlessly support the running of most Windows games. At the Microsoft Build conference held last month, Qualcomm and Microsoft also demonstrated a new development kit to help developers quickly migrate applications to the Arm environment. More importantly, Microsoft's Copilot+ AI has also been integrated with Qualcomm Snapdragon X Elite processors.

Arm takes half of the market share in five years

In the era of AI PC, can Arm compete with X86? At least the Arm camp is confident about this. Arm CEO Rene Hass publicly stated at the Taipei Computer Show that it is expected that Arm architecture chips will get more than half of the market share in Windows PCs in the next five years.

Qualcomm CEO Cristiano Amon also gave a similar optimistic expectation. Amon said that this goal is realistic and feasible. Different OEMs have different goals. Some OEMs even believe that (the proportion of Arm chips) will reach 40% to 60% of their total sales in the next three years, while some believe it will be around 50%. He also emphasized that Windows is transforming towards the AI direction of Copilot+, which will improve productivity and bring new users.

Many OEMs are indeed confident in the growth prospects of Arm chips in Windows notebooks. Dell demonstrated the XPS 13 notebook based on Qualcomm Snapdragon X Elite chip at the Taipei Computer Show, saying that it has increased the battery life of Intel Meteor Lake version from 18 hours to 27 hours.

Industry analysts have also seen the transformation trend of PC chips in the AI era. Daniel Newman, CEO of Futurum Group, believes that "Microsoft publicly announced at the Build conference that Snapdragon X Elite and Qualcomm chip products are the first to meet the Copilot+ standard. This is a huge turning point."

Intel also cannot ignore the shadow of Nvidia. Market rumors say that Nvidia plans to cooperate with MediaTek to launch an Arm-based SoC next year and join the PC chip battlefield in the AI era. Huang Renxun was asked this question when he participated in Dell's event, and he chose a meaningful noncommittal. This gesture gave the market enough hints and expectations.

Intel wants to be a foundry for Arm

Will Intel's dominance in the PC market be shaken in the AI era? This is one of the biggest questions about Intel's future. At the Taipei Computer Show, Intel CEO Gelsinger was constantly asked about the impact of Arm chips, and such questions made him look a little embarrassed.

At least on the surface, Gelsinger is still confident about Intel's position in the PC industry. "This is not the first time there are Windows on Arm products. The market share of x86 is still very hot, and you need a reason to switch platforms. Lunar Lake has the best CPU, GPU and NPU, and longer battery life. I don't think this advantage will be shaken."

However, Gelsinger changed the topic again and mentioned that Intel is willing to manufacture chips for the Arm camp. "We are serious. The cooperation between Intel and Arm is beneficial to both parties, and the benefits will exceed my imagination when I took over as CEO. We see a lot of potential for Arm as Intel's chip foundry partner."

On the one hand, it wants to compete with Arm, and on the other hand, it wants to take orders from Arm. Chip foundry is also a strategic transformation that Gelsinger brought to Intel. Since taking office in 2021, Kissinger has launched the IDM 2.0 strategy, stepped up the pace of advanced process technology, and restarted the foundry business, planning to surpass Samsung in 2030 and become the world's second largest chip foundry after TSMC. Kissinger said in February this year that the expected transaction value of Intel's foundry business, including wafer manufacturing and advanced packaging, has exceeded US$15 billion. In the last fiscal quarter, Intel also announced that the foundry business would be independent as the IFS business unit. In the first quarter of this year, IFS's business revenue was US$4.4 billion, a year-on-year decline of 10%, and an operating loss of US$2.5 billion. However, TrendForce data shows that Intel's foundry business is currently not even ranked in the top ten in the world, with TSMC and Samsung occupying an absolute advantage with 61% and 11.3% respectively. How feasible is Kissinger's goal of making Intel the world's second largest chip foundry in 2030?

The road to reviving Intel is long and difficult

This time, Intel even handed over its flagship chips to TSMC for foundry. The Lunar Lake released by Kissinger in Taipei this time will be manufactured by TSMC's N3B (3 nanometers). This is also the first time that Intel has handed over its flagship CPU to an external foundry (the packaging is still done by Intel itself). Kissinger admitted that the reason why Lanar Lake was handed over to TSMC for foundry was because the latter had better process technology, so he publicly thanked TSMC. However, Kissinger emphasized that the next generation of Panther Lake will be produced by Intel's own process again. At that time, Intel will use the 18A 3nm process to let the outside world see the true strength of Intel's wafer factory. In 2021, Intel, which was in a state of internal and external troubles, invited Kissinger back from VMWare as CEO, hoping that this senior executive who has grown up in Intel for 30 years can reinject Intel's engineering culture and lead Intel to lead industry innovation again. Intel even offered Kissinger a sky-high salary of $178 million, which was even voted down by the shareholders' meeting. But it needs to be explained that Kissinger's $178 million salary includes: $1 million in base salary, $1.75 million in cash bonus, $5.1 million in non-equity incentive bonus, more than $140 million in stock grants, and $29.1 million in options. Among them, options and stock grants are linked to performance and stock price growth. Judging from the subsequent performance, he did not get this large income.

In the three and a half years since Kissinger took office, Intel did not get rid of the industry's difficulties, but witnessed the rapid growth of Nvidia and AMD. From January 2021 to date, Intel's stock price has not climbed, but has fallen by 46%; revenue has also dropped sharply from $77.86 billion in 2020 to $54.2 billion last year. Although Intel has stopped the sharp decline in revenue this year, it is still difficult to recover.

Although Kissinger demonstrated the performance strength of Intel's next-generation chips in Taipei, trying to prove that Intel can still maintain its market position in the AI era, it seems that the market has not been moved. Intel's stock price has hardly changed this week, and its current market value has even fallen below $130 billion.

View more at EASELINK