NVIDIA to Acquire Arm for $40 Billion

On the 13th, Nvidia announced the final agreement to acquire ARM from Softbank. The acquisition amount is 40 billion dollars, the largest in the history of mergers and acquisitions in the global semiconductor industry.

What kind of company is ARM

ARM is somewhat unfamiliar to the public, but is the world’s largest semiconductor design company with a unique position in the Systems-on-a-chip (SoC) market. ARM does not produce semiconductors directly, but designs semiconductor architectures and provides them to semiconductor manufacturing companies. To compare to the food industry, you don’t make and sell food yourself, but you sell recipes for the food and make income whenever a recipe is used. The semiconductor architecture designed by ARM is used in a variety of devices such as smart TVs, drones, and tablets, but is most commonly used for manufacturing smartphone mobile applications (AP). From Samsung’s Galaxy phone to Apple’s iPhone, 90% of all smartphones in the world use ARM architecture. Although ARM generated more than $1.7 billion in sales in 2019, it has recorded a small but continuous operating loss as the growth of the smartphone market slows.

Intel is the representative competitor of ARM. Intel is a traditional powerhouse in the semiconductor industry that not only designs but also manufactures, and has a CPU architecture known as x86. The biggest difference between ARM and Intel’s CPU architecture is the power consumption of the chip. x86 is mainly used in notebooks and desktops that consume a lot of power, and the ARM architecture is mainly used in smartphones because it has a lower ability to perform tasks than x86 but consumes much less power. Intel’s x86 CPU’s global market share in the second quarter of 2020 was 65%.

Nvidia is now a data center enterprise

When you think of a company called Nvidia, computer games will come to mind first. A graphics card is required to run games that require high specifications, because GPU, the core element of the graphics card, is Nvidia’s flagship item. However, Nvidia is not a company that develops only the GPUs needed for computer games. Nvidia is also heavily involved in the data center business, with data center sales rising to $830 million in 2016 and $3 billion last year. In April, it acquired Mellanox for $6.9 billion, which makes hardware that transfers data between chips in a cloud computing system.

A data center is a facility that houses numerous networked computers working together to process, store, and share data. In the past, personal computers have used CPUs, but now data centers will gradually perform computing operations through cloud servers. In the data center, a lot of tasks such as analysis and storage of big data, cloud computing, IoT implementation, and machine learning have to be done at once, and a GPU optimized for training artificial intelligence and deep learning models is more suitable than a CPU. The GPU has a large number of cores that help process parallel processing, so it can process multiple calculations at the same time compared to the CPU, as well as the memory bandwidth that can process a large amount of data, making it suitable for deep learning calculations.

An example of an application in a data center is autonomous driving. Autonomous driving is closely related to data centers because it generates very large amounts of data, ranging from 5 TB to 20 TB per vehicle per day. Nvidia is also developing platform chips for autonomous driving using data centers. Nvidia’s market value rose to $320 billion from $12.3 billion five years ago, showing a very high growth rate.

Nvidia’s Vision

As 5G technology is introduced and data transmission at high speed becomes possible, the role of the data center is expected to increase. As a result, the proportion of Nvidia’s data center business is growing, and investments are increasing accordingly. The goal of the acquisition of ARM, investing $40 billion, is to apply ARM’s low-power chip design to Nvidia’s GPU to create a more efficient chip. Jensen Huang, CEO of Nvidia, said, “Energy efficiency is directly linked to the cost of computing capacity, computing throughput, and provisioning services.” In addition, Nvidia’s technology will be used in all fields of artificial intelligence, from autonomous vehicles to smart meters. Contributing to this is the ultimate goal of Nvidia.

Considerations and gateways for acquisition

The biggest concern about this acquisition is that Nvidia’s rivals Apple, Intel, Samsung, Qualcomm, Broadcom, and Huawei hold licenses for ARM. This is because if Nvidia operates exclusively with the ARM license, it may no longer sell it to competitors or raise royalties. However, Nvidia said that ARM will continue to operate an open license model while maintaining global customer neutrality.

Approval of each authority is a gateway to be overcome. It requires approval from Chinese, British, European Union and US authorities, and could take up to 18 months, he said. In particular, at a time when the trade war between the US and China is intensifying, the acquisition may be impossible if China does not agree.

Nvidia said it was prepared for a conflict with China and said it was confident it would get approval for the acquisition.