Skip to content
  • KOSPI 3254.42 +4.21 +0.13%
  • KOSDAQ 1055.50 +5.25 +0.50%
  • KOSPI200 431.55 +0.17 +0.04%
  • USD/KRW 1151 0.40 0.03%
  • JPY100/KRW 1,044.99 1.69 0.16%
  • EUR/KRW 1,355.13 -1.60 -0.12%
  • CNH/KRW 177.89 0.30 0.17%
View Market Snapshot

AI semiconductors

AI chip market heats up, to tip over $100 bn in deals

By Oct 22, 2020 (Gmt+09:00)

The global semiconductor industry has been swamped with a series of massive deals this year, likely to tip over $100 billion in deal volume as companies ramp up their artificial intelligence (AI)-related semiconductor operations.

Global chipmakers including Samsung Electronics Co. are pouring in as much as tens of billions of dollars into mergers and acquisitions, along with research and development to secure the leading position in the AI semiconductor market, which is estimated to reach 130 trillion won ($118 billion) by 2030, a drastic leap from the projected $18.4 billion this year.

According to semiconductor market researcher IC Insights, deals within the global semiconductor industry have reached around $72 billion so far this year. This will tip over $100 billion once processor manufacturer Advanced Micro Devices (AMD)’s $30 billion acquisition of Xilinx is finalized.

AI chip market heats up, to tip over 0 bn in deals

Some of the notable deals from this year include: SK Hynix Inc.'s $9 billion acquisition of Intel Corp.’s NAND memory chip business; US-based analog semiconductor company Analog Devices' $20 billion acquisition of industry peer Maxim Integrated Products; and US-based graphics chip heavyweight Nvidia's $40 billion acquisition of UK-based chip designer ARM, alongside a number of industry partnerships.

The AI semiconductor sector is a running theme in ongoing and finalized semiconductor deals this year. AI semiconductor technology provides large-scale learning, reasoning, and calculation skills required to execute AI services.

Semiconductor companies are keenly competing for the lead in the early AI semiconductor market, resulting in aggressive M&As and partnerships to foster and complement their AI operations.

For example, AMD is on par with Intel in the CPU market for general PCs but accounts for only 2% market share of the data center CPU market.

Data centers' core competitiveness relies on the ability to process data via AI, which has driven AMD to acquire Xilinx as its core product is the AI accelerator which helps speed up massive data processing.

Also, Nvidia's acquisition of ARM is backed by the company's push to create synergy in the AI semiconductor industry. Industry watchers predict Nvidia to utilize AI technology to seize the Internet of Things (IoT) or autonomous driving market.

“The combination of Nvidia’s AI, graphic technology with ARM's ecosystem will turbocharge ARM's R&D capacity and expand its IP portfolio with Nvidia's world-leading GPU and AI technology,” said Jensen Huang, founder and chief executive officer of Nvidia.

KOREAN SEMICONDUCTOR COMPANIES JOIN AI RACE

Korean semiconductor companies are also joining the race to develop AI semiconductors and to boost their performances. Samsung Electronics’ US-based R&D centers regularly hire seasoned developers for the neural processing unit (NPU), part of AI semiconductor umbrellas.

The NPU has the most optimized semiconductor technology to faciliate deep learning based on artificial neural networks. Currently the NPU is used in AI-based systems, such video and voice recognition features on smartphones. For example, it optimizes camera settings such as exposure and lighting by recognizing the surroundings.

Samsung Electronics has been expanding its AI semiconductor business on the grounds that the scope of the NPU will broaden to autonomous vehicles.

SK Hynix is set to jump on the AI semiconductor industry trend through the acquisition of Intel's solid state drive (SSD) division. The emphasis on data centers’ AI features will result in increased demand for SSDs as they offers speedy data processing.

Write to Jeong-soo Hwang at hjs@hankyung.com

Danbee Lee edited this article.

Comment 0

0/300