An acquisition is reshaping the global chip industry landscape. More precisely, this acquisition is prompting almost all participants in the chip supply chain to reassess their positions and attempt to rebuild momentum.
On September 14, SoftBank and NVIDIA officially announced that they had reached a final agreement for NVIDIA to acquire chip design company ARM for up to $40 billion. This acquisition price is $8 billion higher than the price SoftBank's Masayoshi Son paid to take ARM private from the Nasdaq stock market four years ago.
However, the biggest difference from four years ago is that SoftBank is an investment fund, while Nvidia is an industry leader, a global leader in GPU (Graphics Processing Unit) and AI chips. Its combination with ARM will reshape the chip industry landscape and its competitive boundaries.
ARM is the world's largest chip IP supplier, with over 95% of smartphones and tablets worldwide using the ARM architecture. Due to its long-standing neutral and open business model, ARM has over 1,000 ecosystem partners globally. To date, ARM licensees have sold 180 billion chips, with companies such as Apple, Samsung, Huawei, and Qualcomm being ARM architecture licensees.
Unlike SoftBank's bet four years ago that ARM technology would become the key to future device connectivity and thus lay out its IoT strategy, Nvidia is clearly more determined to standardize its AI technology by combining its own GPUs with ARM CPUs, thereby creating absolute influence in the AI era.
This business case study focuses on this industry turning point, illustrating this silent war from the perspectives of industry, market, and legal dynamics, the power struggles among various parties, and the possible variables and directions.
The technological battle is strangling the throat of the Internet of Things and AI era.
"Don't ask us, ask Qualcomm. They should be more worried." One interviewee bluntly stated the impact of Nvidia's acquisition of ARM on the chip industry.
Why did Nvidia's acquisition of ARM have such a huge impact on the industry?
The answer was already revealed four years ago when SoftBank acquired ARM, because ARM technology points directly to the future era—the Internet of Things world.
ARM, formerly known as ICON Computers, was founded in Cambridge, England in 1978. In 1985, ICON Computers developed a new processor using a reduced instruction set, named ARM. Although ARM was later spun off into an independent subsidiary due to ICON Computers' financial difficulties, it has long adhered to an open and independent policy, licensing its semiconductor IP to any company that needs it. It is known as the "Switzerland of the global semiconductor industry," thus building its dominant position in the mobile processor field. For example, chips that once dominated the industry, such as the Apple A11 and Qualcomm Snapdragon 835, were designed based on the ARM instruction set.
Data shows that 90% of smartphone processors and other types of mobile chips worldwide use chip designs from ARM.
Seeing that ARM technology would become the key to future device connectivity, Masayoshi Son took ARM private in 2016 for $32 billion (a 40% premium) and delisted it from Nasdaq.
In fact, ARM, as an independent company, is not naive. They are optimistic about the cloud capabilities that Masayoshi Son built through the acquisition of US carriers, and both parties agree that promoting ARM's technology through cooperation will generate huge business opportunities in the next 15 to 20 years. For example, Masayoshi Son gave a keynote speech at ARM's TechCon conference in the United States, claiming that the Internet of Things (IoT) will bring about 1 trillion IoT devices, just like the Cambrian explosion.
However, this acquisition failed to achieve its intended goals, as SoftBank had to reluctantly divest itself due to its own financial problems and the failure of several investment projects. Of course, from ARM's perspective, its four years under SoftBank's umbrella also failed to realize its desired further commercialization goals.
Li Feng (pseudonym), a chip industry professional, told the China Business Journal, "The fastest-growing chips in recent years have mainly focused on neural network processors and data centers. Qualcomm, Intel, and Apple have captured the majority of the market share. ARM still needs to strengthen its commercialization efforts." In Li Feng's view, "Many companies use ARM licenses, but ARM doesn't earn much money. In the chip industry chain, most of the profits are still made by Qualcomm, Apple, and others."
This is closely related to ARM's revenue model. SoftBank's financial report shows that ARM's revenue mainly consists of two parts: one is the upfront licensing fee for IP licensing, including the ARM architecture and ARM IP; the other is the royalties collected based on a percentage of the selling price of each chip.
Public data shows that ARM's revenue in 2017, 2018, and 2019 was $1.831 billion, $1.836 billion, and $1.898 billion, respectively. This revenue scale is not only far behind Apple and Qualcomm's revenues of hundreds of billions of dollars, but also, once the high R&D expenses (accounting for 40% of total revenue) are deducted, ARM's profit is not optimistic; in 2019, ARM's net profit was only $276 million.
More importantly, this company, with its limitless technological potential, has shown signs of stagnation in revenue growth, a situation that clearly cannot satisfy its management. This makes Nvidia's emergence both accidental and inevitable.
In response to the questions raised by the acquisition, NVIDIA founder and CEO Jensen Huang stated in a letter to employees, "We will continue ARM's open licensing model, maintain its customer neutrality, serve customers across various industries globally, and further expand ARM's IP licensing portfolio by leveraging NVIDIA's world-leading GPU and AI technologies." Clearly, increasing ARM's bargaining power through such an IP licensing portfolio is a sound strategy. Moreover, the collaboration between the two technology giants in GPUs and CPUs will further solidify their significant influence in the AI era.
“In the newly added computing scenarios, Intel products are no longer available. For example, edge computing is likely to be Nvidia's market; the Internet of Things (IoT) is more likely to be dominated by ARM; and mobile phone chips are dominated by Qualcomm and Apple. The future business opportunities for ARM's technology are definitely in the Internet of Things. Therefore, Nvidia's acquisition of ARM is aimed at the Internet of Things, allowing more mobile devices to install ARM chips,” Li Feng told reporters.
In fact, as of 2020, ARM partners had shipped over 160 billion ARM-based chips. Many manufacturers, including Intel, Nvidia, Apple, Qualcomm, and Samsung, use the ARM architecture in their A-series processors. Nvidia, founded in 1993, has rapidly expanded its business and made significant strides in the field of artificial intelligence, thanks to its GPU solutions being better suited for the parallel computing required for deep learning and its established industry ecosystem.
Nvidia's Chief Financial Officer, Colette M. Kress, stated that the growth stems from the expansion of its AI product portfolio and continued shipments. Nvidia's chips are increasingly being used in devices beyond computers, gradually becoming the new core of AI servers.
However, this development model has also brought Nvidia and Intel into direct competition. Originally operating in different fields, both companies are now entering the field of artificial intelligence, leading them to acquire other companies to supplement their businesses. The significance of ARM to Nvidia in this battle between giants is particularly noteworthy.
"Regardless of whether this acquisition originated from competition with Intel, its actual impact is enormous, whether in terms of technological competition, industrial competition, or the more complex context of inter-national competition. Many companies developing chips based on the ARM architecture will be looking for alternatives," an analyst said.
Market competition and complex multilateral games
The year 2010 holds a special place in ARM's history. Apple offered $8 billion to acquire ARM, but this was declined by then-CEO Warren East. This is because ARM is an independent company that only designs and develops chips and does not manufacture them directly.
But it is precisely this that makes ARM stand out in the industry, and more and more chip manufacturers are confident in continuing to purchase its IP licenses. Now, Nvidia's acquisition is disrupting this trust and balance.
This directly attracted the first opponent of the acquisition – ARM co-founder Hermann Hauser. In his “letter of appeal” to the British Prime Minister, Hauser stated that “selling ARM to Nvidia would affect the jobs of thousands of ARM employees in Cambridge, destroy the very foundation of ARM’s business model, cause ARM to lose its ‘Swiss’ status in the semiconductor industry, and make it difficult for ARM to remain neutral in commercial competition.”
Therefore, it is called for the UK government to help ARM IPO on the London Stock Exchange and make it a British company if Nvidia cannot simultaneously meet the three conditions of retaining the Cambridge job, not enjoying preferential treatment over other ARM customers, and the UK obtaining an exemption from US CFIUS regulations.
In fact, not only Hermann Hauser, but many market analysts and industry researchers have also written articles stating that "the acquisition will disrupt ARM's business model and create more competitors for Nvidia itself."
"Independence is the most important value of ARM and its ecosystem. Once acquired by Nvidia, this independence will be questioned no matter what guarantees and promises Nvidia makes," a British market researcher wrote in a blog post.
In fact, reports indicate that many major manufacturers are already looking for alternatives or replacements for ARM. For example, Samsung and Apple began developing their own chip architectures as early as 2013; Qualcomm's Snapdragon 820 also returned to its own architecture. Huawei has successfully joined the ranks of ARM's competitors—the RISC-V chip architecture. This architecture, like Android, is open-source and has been welcomed by many Chinese manufacturers.
Clearly, it is widely believed that if Nvidia cannot turn its promises into "legal provisions," it will be unable to withstand pressure from the U.S. national level, because this is not just a competition in industry and technology, but also involves geopolitical and national rivalry.
This has complicated and made the future of the acquisition uncertain. For example, in the UK, on the one hand, a British government spokesperson said that the UK was "closely monitoring the proposed acquisition," but on the other hand, its Shadow Business Secretary Ed Miliband accused the British government: "This company is going to be swallowed up by Nvidia, and the British government is doing nothing."
In China, ARM and TSMC are the two companies with the greatest influence on the Chinese chip industry chain. TSMC almost monopolizes chip manufacturing, while ARM's technology licensing dominates the world of mobile chips.
Public data shows that there are as many as 150 Chinese customers who have licensed ARM architecture, including Huawei, which has obtained a permanent license for ARMv8. Approximately 95% of high-end chips designed in China are based on ARM technology, and the Chinese market contributes about 20% of ARM's sales.
Given that the vast majority of chip designs in my country are based on the ARM reference architecture, once Nvidia's acquisition of ARM is completed, it will be difficult for ARM to remain neutral. It will inevitably become a "Sword of Damocles" hanging over the heads of Chinese chip design companies, subject to the constraints of the United States' long-arm jurisdiction, just like the Huawei "choke point incident" this time, and be restricted by others on a larger scale.
In response, on the one hand, Chinese chip design companies are further increasing their research and development on the RISC-V architecture, ultimately aiming to break free from the limitations of the ARM reference architecture. On the other hand, they are also seeking third-party solutions.
Back in 2016, to address the issue of ARM exporting its core technologies to China, a fund institution with local government backing and ARM jointly established Arm Technology (China) Co., Ltd. (hereinafter referred to as "ARM China"). According to Tianyancha data, Arm China was established on December 21, 2016, with a registered capital of US$66,104,944.14. ARM Limited, the suspected actual controller, holds a 47.33% controlling stake in Arm China.
In addition, Amber Leading (HONGKONG) holds 36% of the shares, Ningbo Meishan Bonded Port Area Anchuang Growth Equity Investment Partnership (Limited Partnership) holds 13.3%, ARMEcosystem Holdings holds 1.7%, TL1016 Technology Limited holds 1.2%, and Ningbo Meishan Bonded Port Area Anchuang Investment Management Partnership (Limited Partnership) holds 0.47%.
Almost simultaneously with the honeymoon period between SoftBank and ARM, in June 2020, ARM China was embroiled in a leadership change controversy. On one hand, major shareholders ARM and Hopu Investment jointly decided at the ARM China board meeting to remove Wu Xiong'ang as chairman and CEO. On the other hand, ARM China objected, arguing that he had violated board meeting procedures and rules of procedure. Several months later, Wu Xiong'ang has remained at the helm of ARM China.
Latest news indicates that, following ARM's dismissal of Wu Xiong'ang, CEO of ARM China, Ningbo Meishan Bonded Port Area Anchuang Investment Management Partnership (Limited Partnership), an investor in ARM China, has recently filed a lawsuit against ARM China in the Shenzhen Qianhai Cooperation Zone People's Court.
All of this makes the antitrust review of Nvidia's acquisition of ARM in China more sensitive and uncertain. The only certainty seems to be that ARM's future market share may fall short of expectations, but this could also affect the outcome of the merger.
According to publicly available information, NVIDIA set profit targets for ARM in this acquisition—SoftBank could only receive $5 billion in cash or NVIDIA stock from the $40 billion "transfer fee" if ARM's operating performance met the agreed targets.
Thus, Nvidia gambled on the future of the technology market, but the risks were left to be borne by SoftBank, which was already overburdened.
Legal game involving uncertain interests of multiple parties
If Nvidia's determination to deeply position itself for the era of artificial intelligence has opened a window and a new turning point for technological competition in the chip industry, and if the acquisition obstacles determined by ARM's business model will greatly influence the market changes in the chip industry chain, then this acquisition of a British company (ARM) by an American company (Nvidia) from a Japanese company (SoftBank) against the backdrop of Sino-US technological competition has pushed the complexity of the antitrust review triggered by this cross-border acquisition to an unprecedented level.
On July 8th, Nvidia surpassed Intel ($249.3 billion) to become the most valuable chip manufacturer in the United States, with a total market capitalization of $251.3 billion. This provides Nvidia with stronger financial resources for external acquisitions, especially stock-swap acquisitions. On the other hand, it will also affect, to some extent, its market position assessment and related calculations in antitrust reviews.
According to lawyers, the acquisition is likely to be subject to antitrust investigations from regulatory authorities in the United States, the United Kingdom, the European Union, and China. In the UK, ARM is considered the only technology company capable of generating global revenue and influence, naturally attracting considerable attention, with many hoping it will remain a British company rather than an American one. For example, Hauser stated that an acquisition of ARM by an American company could lead to the UK losing its "economic sovereignty."
Furthermore, the EU's antitrust review may carry a stronger "negative" connotation. This stems from the chip industry strategies of EU member states such as France, Germany, and the Netherlands. In addition, after the US cut off supplies to Huawei, some European companies proposed supplying chips that do not use US technology, targeting China's $300 billion chip import market.
"Based on the information released so far, it's unknown whether they used the ARM design framework. However, restrictions on ARM will inevitably affect the flexibility of the solutions offered, which may also influence the EU's rulings on antitrust issues to some extent. In fact, judging from the antitrust investigations and various legislations introduced by the EU in recent years, the EU has been striving to find a balance with the US in the technology industry competition. Given the enormous influence of the chip industry on future technological development, it's impossible for the EU to remain inactive on this issue," a relevant lawyer told reporters.
Data shows that the European Commission is one of the world's most stringent competition enforcement agencies in regulating tech giants, having initiated a total of 19 antitrust investigations against the four major US tech giants. On October 16, 2019, the EU announced provisional antitrust measures against Broadcom.
In China, this cross-border merger and acquisition would also require approval from the Chinese Ministry of Commerce.
Clearly, whether from the perspective of market share, or from the perspective of national welfare and the interests of domestic investors, this merger and acquisition will be difficult to bypass the scrutiny of regulatory authorities.
Undoubtedly, beyond competition in technology, industry, and markets, this acquisition will also spark a global battle of legal expertise.
Observing the entropy of the chip industry development
In the centuries since German physicist Rudolf Clausius proposed the concept of "entropy" in 1865, people have gradually come to realize that everything in the universe tends from order to disorder, from balance to chaos and uncertainty, but will eventually rebuild balance.
The current chip industry is similarly in a state of transition from order to disorder, awaiting the restoration of balance. As some analysts have pointed out, "Nvidia's acquisition of ARM has merely opened a crack in the changing landscape of the chip industry."
A technical director of a well-known domestic mobile phone manufacturer told reporters: "Regardless of whether or not this acquisition occurs, the integration of CPUs and GPUs is definitely a trend. Apple's A13 and A14 chips are already undergoing integration. In particular, the demand for GPUs in the mobile phone business is increasing, which will readjust the positions of different companies in the chip industry chain."
Behind this integration, collaboration will become increasingly important. The most typical example is the increase in the number of acquisitions. For instance, Intel has been rapidly expanding in the field of artificial intelligence in recent years by acquiring startups, and has even marked GPU solutions in its roadmap.
Similarly, in 2019, Nvidia beat Intel to a deal to acquire the Israeli company Mellanox for $6.9 billion, marking the largest acquisition in Nvidia's history.
Moreover, cooperation within the chip industry chain has a deeper meaning. The chip industry comprises a vast and complex supply chain, broadly divided into four main segments: design, manufacturing, packaging, and testing. Each segment requires seamless integration and high levels of interaction. The manufacturing segment alone necessitates the concentration of the world's most advantageous resources. Take ASML, the global leader in lithography machines, as an example. Its EUV lithography machines, due to their extremely high manufacturing complexity, require collaboration from top companies across multiple countries and sectors, representing the pinnacle of industrial manufacturing achievements.
Or perhaps this is precisely the opportunity that Masayoshi Son sees. In his view, with chips at the core, it is entirely possible to build a globally connected industrial ecosystem based on the era of the Internet of Things.
However, Masayoshi Son misjudged the development of the international situation. When the element of the nation was highlighted, the originally perfectly designed "globalized industrial ecosystem" had to be fragmented due to the technological competition between countries (which sometimes cannot avoid geopolitical relationships).
The US's CFIUS and high-tech export control laws have first and foremost strangled Chinese tech giant Huawei, and have inevitably impacted the revenue of upstream and downstream companies in the industry chain. It is well known that the chip industry has a long R&D cycle and requires huge investments. Once a product is successful, a large portion of the market will be focused on China, the world's largest consumer market; without consumption, there is no revenue. From this perspective, this is a major blow to the entire global chip industry ecosystem.
According to public reports, Samsung, SK Hynix, TSMC, MediaTek, and other manufacturers have submitted applications to the U.S. Department of Commerce, hoping to continue supplying products to Huawei. Qualcomm, on the other hand, is attempting to lobby Trump to lift restrictions on selling chips to Huawei, otherwise it risks handing over a market worth up to $8 billion to Qualcomm's overseas competitors.
As a result, the chip industry chain is facing the biggest paradox in its history, both in terms of technology and market. The true meaning of the industry chain requires global cooperation and linkage, while the impulse of technological competition among countries will continue to disrupt the original balance.
