In September 2020, the announcement of NVIDIA’s takeover of Arm Holdings in a transaction valued at $40 billion created both excitement and concern. What does this deal imply for the future of AI, Edge computing and the overall semiconductor industry? Will the deal enable NVIDIA to become the next tech monopoly besides companies such as Facebook and Google? Even if these questions remain unresolved, the deal is expected to influence the semiconductor industry and its players significantly.
The semiconductor industry is highly complex and globally dispersed. It comprises companies that engage in the design and precision manufacturing of semiconductors. Semiconductors can be found in millions of digital devices and are critical to the function of everyday consumer electronics, computing devices and communications. Thus, they encounter us in almost every aspect of our lives and are relevant to all economic sectors. In 2019, the size of the global semiconductor market amounted to $513 billion and is expected to reach $726 billion by 2027, representing a compound annual growth rate (CAGR) of 4.7% over the forecast period. This growth is attributed to the increasing global demand for consumer electronics devices, electric vehicles, integrated circuits from emerging countries and memory devices. Additional trends in the semiconductor industry evolve around Machine Learning, Artificial Intelligence and Internet of Things (IoT), which provide new opportunities for market development. In general, the semiconductor industry is highly cyclical with its financial performance following the economic patterns. Consequently, the outbreak of COVID-19 led to several challenges affecting the capacity of the industry to maintain stability. Future development depends on the economy’s ability to cope with the global pandemic.
Key players in the semiconductor industry include INTC, TSM, QCOM, Texas Instruments, AMD, NVIDIA and ASX. These companies strive to produce cheaper, smaller and faster semiconductors to facilitate increasingly affordable, streamlined and powerful technological devices. Generally, semiconductor companies apply different levels of specialization and functional delineation in the value chain resulting in two key operating models in the industry: integrated device manufacturer (IDM) and Fabless-Foundry. In the IDM model, one company performs all production phases from design, manufacturing, assembly and testing to packaging. In the Fabless-Foundry model, production phases are divided: Design companies focus on design and outsource precision manufacturing, consequently being “Fabless.” Foundry companies focus on contract manufacturing. The third group of companies operates assembly, testing and packaging. The latter group is renowned as outsourced semiconductor assemblies and test companies. In recent years, organizations are steadily adopting characteristics of the Fabless-Foundry model as it is highly cost-intensive to remain in pace with technological progress.
NVIDIA operates worldwide as a visual computing company, designing graphics processing units and System on a Chip (SoC) units. The Group, headquartered in Santa Clara, California, was founded in 1993 by Jensen Huang (NVIDIA’s current CEO), Chris Malachowsky and Curtis Priem. NVIDIA’s success story begins with Graphics Processing Units (GPUs) intervention in 1999. Subsequently, NVIDIA invented GeForce 256, the first GPU, putting it on a path to reshape the industry. In the same year, NVIDIA announced its initial public offering on the NASDAQ Stock Exchange at $12 per share.
NVIDIA identifies four different core businesses within the broader segments it operates: Gaming, Data Centers, Professional Visualizations and Automotive. Gaming is NVIDIA’s largest business generating an expected $5.5bn in revenues for FY2020 and therewith representing 51% of its total revenues. NVIDIA has a strong market positioning in the Gaming business being the technology leader with its main product called GeForce, the world’s largest gaming platform (+200m gamers worldwide). NVIDIA’s second core business by revenues is Data Center, where the main products address High Performance Computing (HPC) and advanced AI technology needs. The Data Center business is expected to generate up to $3.0bn in revenues for FY2020 representing 27% of its total revenues. Professional Visualization is NVIDIA’s third core business area providing solutions with its cutting-edge graphics workstations to designers and creatives. In this market, NVIDIA holds more than 90% market share leading to expected revenues of $1.2bn for FY2020 i.e. 11% of its total revenues. NVIDIA’s final core business, Automotive, is expected to generate $0.7bn in revenues for FY2020. This business area is mainly focused on offering solutions to infotainment and autonomous vehicle manufacturers. NVIDIA counts on 223 high-profile drive partners, such as Mercedes-Benz, Toyota, Volvo, Continental and DiDi.
Overall, NVIDIA operates in the broad semiconductor industry. NVIDIA generates a total of $10.9bn revenues for FY2020 with an Adjusted EBITDA of $4.1bn, leading to an Adjusted EBITDA Margin of 37.6%. Last year, NVIDIA saw a top line reduction of approximately $0.8bn, mainly due to a contraction in its Gaming business. NVIDIA’s stock price equals $537.61 (as of 11/19/2020) and broker consensus estimates a 1-year target price of $567.50. Consequently, NVIDIA’s current market capitalization equals $323.4bn (as of 11/19/2020).
Arm is a semiconductor and software design company focusing on the design and development of computer processors (CPUs), software development tools, cloud services and IoT solutions. Founded in 1990, the company, headquartered in Cambridge, UK, was listed on the London Stock Exchange and NASDAQ in 1998. In September 2016, Arm was acquired by SoftBank for a total acquisition price of $31.0bn, including a 43% premium to the market for “a company that could be worth more than Google”, commenting Masayoshi Son, founder of SoftBank. After the completion of the acquisition, Arm was delisted from the London Stock Exchange and ceased to be a listed company. Building on the estimation that every person uses an average of 20 Arm-designed chips, Arm can be considered as one of the world’s most valuable intellectual property companies issuing licenses to +500 companies, ranging from Apple to Huawei.
Arm’s business is primarily focused on the licensing of intellectual property. Hence, Arm licenses processor designs to semiconductor manufacturers that incorporate Arm’s technology into their chips. Semiconductor manufacturers pay an up-front fee to gain access to the technology and a royalty on the chips produced by Arm, typically related to the chip’s price. Arm operates in 5 different reference markets:
- Mobile application processors
Arm has a market share of more than 90% in FY2019, and the market is developing at a fast pace creating further opportunities for Arm. The company expects to consolidate its market share in the next 10 years
- Networking equipment
Networking infrastructure will become more scalable and energy efficient in the coming years and Arm expects to double its market share within 10 years from 32% to 65%
- Data Centers
Arm currently has a market share of 5% in this reference market and intends to increase it to +25% by 2028
- IoT application processors
Arm is a market leader for IoT processors holding a market share of 90% and expecting to consolidate it in the next 10 years
- Infotainment and driver assistance
Arm holds a market share of 75% and is committed to investing in order to gain a 90% market share in this reference market within the next 10 years
Generally, Arm generated Net Sales of $1.9bn in FY2019 representing an increase of 3.4% compared to FY2018. Arm’s revenues are accounted for and divided into technology royalty revenues, technology licensing revenues and software and services revenues that account for $1.1bn, $582m and $235m, respectively.
On the 13th of September 2020, it was announced that NVIDIA agreed to acquire Arm from SoftBank. Rumors of the deal were reported already in July and, despite the high interest of other companies such as Samsung to acquire a stake of Arm, NVIDIA succeeded in buying 100% of the company. The deal is expected to be completed on the 13th of March, 2022.
Arm was sold by SoftBank (75.01%) and SVF Holdco (24.99%) and the deal value amounts to $40bn. The specifics of the deal are detailed below:
- $12bn paid in cash ($2bn to Arm and $10bn to SoftBank)
- $21.5bn paid to SoftBank in shares
- In case of reaching certain financial performance targets, NVIDIA will pay an additional earnout of up to $5bn in either cash or shares
- $1.5bn in stock awards will be received by Arm’s employees for post-closure retention
NVIDIA’s share price has skyrocketed during the past year. Consequently, NVIDIA would benefit from financing the deal with stocks. NVIDIA finances the acquisition through a capital increase and vendor placing. After the deal, Softbank will become NVIDIA’s largest shareholder with a stake between 6.7 and 8.1%.
The deal has been part of the tech and telecoms divestiture strategy that SoftBank has undertaken this year. At the peak of the COVID-19 pandemic, Softbank launched a program to cover its losses, fund share buybacks and reduce debt exposure for a total value of $40bn. Selling its stakes in Alibaba, T-Mobile and Arm, Softbank exceeded its provisions.
Yet, this wasn’t the first time SoftBank and NVIDIA got involved in negotiations. Already, back in 2016, when SoftBank was considering investing in the semiconductor industry, both Arm and NVIDIA represented interesting options. SoftBank picked the former, missing a chance to ride the rally of what is today the world’s most valuable chipmaker. As depicted in Figure 1, SoftBank bought its stake in Arm in 2016 for $32bn making only a small gain on its investment. Generally, SoftBank is one of the most discussed companies in the world, and this investment left many doubts about its stock-picking decision.
Figure 1: Interconnection of Arm, NVIDIA and SoftBank
NVIDIA’s acquisition of Arm combining NVIDIA’s leading AI computing platform with Arm’s vast CPU ecosystem reaching up to 70% of the world’s population leads to the creation of a premier computing company. Therewith, NVIDIA’s intention to increase its market power by becoming the main supplier of core technologies powering large-scale data centers can be identified as the main objective of the acquisition. Furthermore, the opportunity to expand Arm’s IP licensing portfolio with NVIDIA’s technology in large end markets is a significant driver of the deal.
In addition, the acquisition of Arm by NVIDIA will result in increased R&D investment for Arm, which will increase from $0.5bn to $3.2bn per year (representing approximately 24% of Arm’s current revenues). The “Turbocharge investments” would enable Arm to accelerate data centers, edge AI and IoT capabilities while increasing the reach of NVIDIA’s computing platform from 2 to over 15 million developers.
From a financial point of view, the deal appears attractive. Arm has a strong financial profile with high profitability (e.g. expressed through an Adjusted EBITDA margin of 35%) and a high-quality business model. The deal is expected to be immediately accretive upon closing.
However, there are also some roadblocks to be assessed. Firstly, Arm’s business model varies greatly from NVIDIA’s resulting in potential challenges for post-merger integration. While Arm is engaged in licensing intellectual property to other companies, NVIDIA focuses on the sale of semiconductors. So far, no company apart from QCOM has succeeded in operating both businesses simultaneously. Another challenge arises from the fact that Arms’ customers include among others, competitors of NVIDIA. Post the completion of the acquisition, these competitors will be dependent on NVIDIA. Yet, as of today, there is no alternative to Arm’s Risc-V for NVIDIA’s competitors. Finally, the British government is reluctant to transfer Arm into US ownership and has concerns about NVIDIA dislocating Arm’s headquarter from Cambridge to California, where NVIDIA is based and committed to the creation of an AI hub. Jen-Hsun Hang, CEO of NVIDIA, attempts to mitigate this concern by promising continued investments in the UK without dislocating Arm’s headquarter.
 Internet of Things: System of interrelated, internet-connected objects that are able to collect and transfer data over a wireless network without human intervention.
 Infotainment: Broadcast material that is intended both to entertain and inform.
References: NVIDIA’s website, Arm’s website, SoftBank’s website, Financial Times, Bloomberg, Reuters, The Guardian, MarketScreener
To contact the authors: