Half a century of struggle: What major changes prompted Intel AMD cooperation?

Although they have been in a fierce competition for half a century, facing uncertainties in the common foundation, these arch-rivals have finally decided to join hands to jointly maintain each other's ecosystem. After all, their past, present, and future are all built on the prosperity of the x86 platform.

Historic Collaboration of Arch-Rivals

Last week, Lenovo Group held its annual Tech World technology innovation conference in Seattle, Washington, USA, showcasing artificial intelligence products and strategies spanning hardware devices, servers, and enterprise solutions. CEOs of six major tech giants including Microsoft, Intel, Meta, AMD, NVIDIA, and Qualcomm took the stage or appeared in videos to support Lenovo, giving Yang Yuanqing face and demonstrating Lenovo's position as a global hardware partner.

In addition to Lenovo's own AI strategy release, the biggest highlight of this conference was undoubtedly the first-time collaboration between old rivals Intel and AMD. Intel CEO Gelsinger and AMD CEO Su officially announced their collaboration, surprising many attendees. The two CEOs even took a historic friendly photo, demonstrating their willingness to cooperate.

What brought these arch-rivals of half a century together? It is, of course, their common business foundation: the x86 platform. These two chip giants of the x86 architecture, along with many chip, software, and hardware partners, have jointly established an x86 ecosystem advisory group, planning to continue maintaining the prosperity of the x86 architecture ecosystem in the AI era.

Intel CEO Gelsinger stated in his keynote speech, "Today may be an unprecedented first collaboration between Intel and AMD. Some say the x86 era is over? I want to tell you, rumors of the 'death' have been greatly exaggerated, we are still alive, and we are doing well, x86 is thriving."

Gelsinger then announced, "We are entering one of the most important periods of innovation for the x86 architecture. As the x86 architecture that has been the foundation of computing for decades is about to undergo a period of customization, expansion, and scalability, this is the opportunity brought by the AI era. Intel and AMD announced the formation of the x86 ecosystem advisory group. There is no better stage than Lenovo's to announce this."

He further explained, "In this era of artificial intelligence, 3D packaging, and chipsets, we believe this will open up innovative categories around systems and new workloads. We welcome Lenovo as a founding member to join this advisory group, and we are honored to see other companies joining Intel, AMD, and Lenovo."

After Gelsinger stepped down, AMD CEO Su took the stage and said, "Our x86 ecosystem advisory group illustrates the uniqueness of the technology field in today's era. Ultimately, the direction we strive for is to accelerate the development of computing capabilities and drive the widespread application of computing technology. The x86 architecture has been leading the industry for the past 40 years, and this initiative by AMD and Intel, together with all founding members, is to accelerate innovation in the future." The photo of the two CEOs was taken against this background. So, what is the x86 Ecosystem Advisory Group going to do? This organization brings together leaders from the entire ecosystem, aiming to shape the future of x86, simplify software development, ensure interoperability and interface consistency, and provide developers with standard architectural tools, instruction sets, and a clear outlook for the future.

In addition to the two chip giants and Lenovo, the founding members of this organization also include Microsoft, Google, Meta, HP, Huawei Technologies, Broadcom, Dell, Oracle, and Red Hat. Furthermore, the founders and CEOs of legendary games, Sweeney, and the founder of Linux, Torvalds, are also founding members.

Half a century of open and secret struggles

Intel joining hands with AMD? This was hard to imagine just a few years ago, as these two chip giants have been arch rivals for the past half century. The intense competition history between the two companies runs through the entire history of the PC era.

Perhaps "arch rivals" is not quite accurate, because in the history of the past half century of the two companies, most of the time Intel has played the dominant role of leading the industry trends in technology, while AMD has been the little brother chasing hard behind, with a significant technological gap with Intel, relying only on cost-effectiveness to gain some market share.

Founded in 1968, Intel's two founders, Robert Noyce and Gordon Moore, left Fairchild Semiconductor to start their own business due to different ideas. They then invited Andy Grove to join, becoming Intel's third employee. Over the next few decades, these three men successively became CEOs, helping Intel become a symbolic company in Silicon Valley.

A year after Intel was founded, another Fairchild Semiconductor employee, Jerry Sanders, also left to found AMD. Sanders did not come from a technical background but from sales. Compared to Intel, which had strong R&D capabilities and led industry trends, AMD's reputation and influence were much smaller, and they could only focus on cost-effective products.

Although they developed the first microprocessor, the Intel 4004, three years after its founding, throughout the 1970s, Intel's main business was still memory chips. It wasn't until the early 1980s, after facing strong competition from Japanese companies, that Intel, under Grove's leadership, decisively transformed, abandoning the memory chip business entirely and turning to microprocessors.

Both companies faced a major turning point in the early 1980s. Under pressure from the U.S. Department of Justice's antitrust lawsuit, IBM decided to abandon its idea of dominating the entire PC ecosystem, handing the operating system to Microsoft and the processor to Intel's 8086 processor, focusing on PC hardware. IBM at the time certainly did not expect that this arrangement would make Microsoft and Intel giants in the software and chip industries, while they themselves would give up the PC hardware business twenty years later.

Although IBM handed over the supply of processors for the PC business to Intel, for the sake of balancing the supply chain, IBM wanted to add another supplier to ensure supply stability, prompting Intel to sign an authorization agreement in 1982, allowing AMD to join the camp of the 8086 processor Several years later, Intel once reneged on authorizing x86, forcing AMD to start a long legal battle. It wasn't until the early 1990s that AMD won the lawsuit and reconfirmed its right to be compatible with x86. After that, Intel did not tear up the authorization agreement again because they had become an industry giant. Due to the risk of antitrust regulation, it was necessary to keep AMD as a competitor.

Over the past forty years, Intel and AMD have been the two major suppliers of PC processors. For most of the time, Intel has held an undisputed dominant position. Processors like 486, Pentium, and Core have directly driven the rapid growth of the PC industry, with a market share exceeding 80% at one point. AMD, on the other hand, has always lived in the shadow of the giant, gaining some market share with cost-effective products.

AMD has also had its moments of glory. With the Athlon K7 and K8 processors at the turn of the century, it once overwhelmed Intel. Not only were AMD's processors cheaper than Intel's Pentium, but they also had higher performance frequencies. In recent years, taking advantage of Intel's 10nm process bottleneck, AMD has made a comprehensive comeback in the PC and server processor fields, pushing Intel into a historic low.

The competition between the two companies is not only in the product field but also involves various non-competitive means. In addition to the x86 architecture authorization lawsuit, AMD also sued Intel for unfair competition in 2005, accusing Intel of using its market dominance to pressure OEM manufacturers through discounts and subsidies to purchase only Intel processors. A few years later, Intel was forced to pay $1.25 billion to settle with AMD in the United States and was fined €10.6 billion by the European Union for antitrust violations.

Intel's frustration

The intense competition between the two companies begs the question: what made them set aside their differences and cooperate? It is the technological changes brought about by the mobile and AI eras. From the PC market to the server market, the Arm architecture has brought a huge impact to the x86 architecture, making Intel the most frustrated chip giant.

Intel was once the flagship company of Silicon Valley and the undisputed largest giant in the global chip industry. But now, this chip giant is in an unprecedented crisis, not only losing the title of the world's largest chip company but also possibly being removed from the Dow Jones Industrial Average component stocks.

Intel's current market value is less than $100 billion, even lower than it was over a decade ago, and far below many of its chip peers. In comparison, Qualcomm has a market value of $190 billion, AMD has a market value of $250 billion, even Arm, which is only involved in design, has a market value of $160 billion. Not to mention Nvidia, leading the AI chip industry, with a market value soaring to $3.3 trillion.

The fundamental reason for Intel's low stock price is its outdated product technology and lack of competitiveness in business. Due to strong competition and market share erosion in the PC and server chip markets, from 2020 to 2023, Intel's revenue plummeted from $77.9 billion to $54.2 billion, and net profit dropped from $20.9 billion to $1.69 billion When Intel was stuck in a chip manufacturing bottleneck, AMD, on the other hand, relied on TSMC's leading technology to be the first to launch 7nm and 5nm processor products, becoming TSMC's second largest customer. From 2020 to 2023, AMD's revenue surged from $9.76 billion to $27.7 billion. According to Statista's data, Intel's share in the x86 CPU market has dropped to 63% in the third quarter of this year, while AMD has gained a 33% share. AMD's EPYC processors have also captured a 24% market share in the server market.

Although AMD's revenue is only half of Intel's, AMD's market value is now two and a half times that of Intel. This is because the capital market actually reflects future growth expectations. Theoretically, AMD could easily acquire Intel while Intel is at a historic low, burdened by heavy financial obligations due to investing heavily in chip factories.

In August of this year, Intel announced a global workforce reduction of 15%, cutting costs by $10 billion, and there were reports of considering strategic choices, possibly selling off parts of its business. According to authoritative U.S. financial media reports, Qualcomm has approached Intel with the intention of acquiring its chip design business, but no agreement has been reached between the two parties.

This news undoubtedly shocked the entire industry, as it not only indicates Qualcomm's desire to swallow Intel but also signifies the victory of Arm architecture over x86 architecture. After dominating the entire smartphone era, Arm architecture processors are now challenging the dominance of x86 processors in PCs and server chips.

x86 Facing Strong Challenges

In 2020, Apple began using its self-developed M-series processors based on Arm architecture, leveraging inherent design advantages for performance, efficiency, and cost leadership. In just two years, Intel completely lost its Mac processor business. Although Apple's share in the global PC market is not large, with only 7.8% in the third quarter of this year according to IDC data, Apple's successful transition has given Qualcomm hope to impact the PC market.

Although Qualcomm introduced laptop processors as early as 2017, performance and compatibility have always been lacking, focusing on network connectivity and long battery life, unable to pose a substantial threat to Intel's x86 processors. The market landscape underwent a major change in 2023. In 2021, Qualcomm acquired NUVIA founded by the chief architect of Apple's chip design team for $1.4 billion, greatly enhancing its chip design team.

In 2023, two years later, Qualcomm officially released the Snapdragon X Elite laptop chip, whose powerful performance and efficiency not only surpassed Intel's Core i9 but even exceeded Apple's custom M3 processor. The Snapdragon X Elite processor also realized on-device generative AI, laying the foundation for the subsequent launch of Copilot+PC in collaboration with Microsoft This year's AI PC wave is the best time for Arm architecture processors to truly impact Windows laptops. Newman, CEO of industry analysis firm Futurum Group, believes that "Microsoft publicly announced at the Build conference that the Snapdragon X Elite and Qualcomm chip products are the first batch to meet the Copilot+ standard, which is a huge industry turning point."

The Arm camp, represented by Qualcomm, is full of confidence. Microsoft executives expect that by the end of this year, shipments of Arm architecture laptops will reach 1-2 million units, with a market share in the Windows camp reaching 5%. Arm CEO Haas even publicly stated at this year's Taipei Computer Show that Arm architecture chips will capture more than half of the market share in Windows PCs in the next five years.

Qualcomm CEO Amon also gave similar optimistic expectations. Amon stated that some OEM manufacturers even believe that they will reach 40-60% of their total sales in the next three years. He emphasized that Windows is transitioning towards AI with Copilot+, which will enhance productivity and bring in new users.

Given the relatively stable PC market shipments, if the market share of Arm architecture PC processors significantly increases, it means a decline in x86 market share, with Intel and AMD being the direct victims. This strong impact from competitors is the direct motivation for Intel and AMD to work together to maintain the x86 ecosystem.

In addition, the RISC-V open-source instruction set architecture is also growing rapidly, not only gradually supported by mainstream operating systems but also welcomed by the eager-to-breakthrough Chinese technology industry. China has become a major promoter of the RISC-V architecture, with half of last year's 10 billion RISC-V architecture core shipments coming from the Chinese market.

The main contributor to the RISC-V architecture, SiFive's co-founder and chief architect Asanovic, predicts that the RISC-V architecture has no performance ceiling or application restrictions, can have better processors and ecosystems, and is expected to surpass traditional architectures in the next three years. Research firm Semico Research predicts that by 2025, RISC-V's annual core shipments could reach as high as 40 billion.

In fact, Intel is also investing in the RISC-V architecture, announcing a $1 billion investment in 2022 and joining the RISC-V International Board of Directors. In addition, Intel has also invested in the RISC-V startup company RIVOS, which specializes in designing RISC-V architecture processors for the server market for AI data workloads.

Collaborating while competing

So, how will Intel and AMD cooperate to maintain the future competitiveness of the x86 camp? Although the specific details of the cooperation have not been announced, it is certain that this newly formed group plans to standardize additions and modifications to the x86 ISA, including some ongoing simplification work For example, AMD has been expanding through the supervisor portal, clearing out the outdated parts of the x86 instruction set architecture, while Intel also has a similar FRED code. Intel has even started developing X86S, a simplified, 64-bit instruction set designed to remove more historical legacy parts of x86.

Cleaning up old code is clearly a potential area for cooperation, but the x86 instruction set still needs to continuously support new expansions, which requires more collaboration between Intel and AMD. For instance, Intel recently introduced the built-in accelerator AMX, a matrix operation extension that can significantly enhance AI inference workload capabilities. In the future, x86 will definitely have more extensions, especially those related to AI operations, all of which will require cooperation between Intel and AMD.

After the two companies announced their collaboration, AMD's Executive Vice President of Data Center Business, Norad, stated, "x86 is the de facto industry standard. It is a powerful ecosystem, although Intel and AMD have jointly developed it in some way, they have kept a certain distance from each other. This has led to some inefficiencies and deviations in the instruction set architecture."

While Intel's Executive Vice President of Data Center and AI Business, Hotard, stated, "We are seeing a real shift in computing demands, and the industry is at a turning point, so this is the best time for collaborative research on new innovations in the x86 architecture."

Better collaboration and standardization will also benefit security-related functions. Standards like PCIe, ACPI, and USB are the results of close collaboration across the industry. Perhaps in the future, Intel and AMD may have disagreements on certain collaborative decisions regarding x86, but the presence of other companies and customers in this group will serve as guiding forces to help ensure that the x86 camp moves in the right direction.

Of course, advancing the x86 ecosystem through collaboration does not mean that the two companies will lay down their arms. Norad from AMD stated, "We will still engage in fierce competition. Hotard and I are friends in private, but when we work, we will give our all for our respective companies to ensure that our customers have attractive product choices. However, we can promote industry standards while competing, and that's what AMD and Intel have been doing all along."