Huawei to popularize 5G-based AI computing platform
SoftBank to Provide 5G to 99% of U.S. Urban Areas
Samsung Collaborates with Fabless on 3GAE Process Technology South Korea declared 2019 as the year of 5G commercialization after it first demonstrated 5G technology at the 2018 Pyeongchang Winter Olympics. SKT President Park Jung-ho explained, “5G is the offline world itself coming wirelessly.” He recognized 5G and AI as a convergence relationship in the era of the 4th Industrial Revolution, and expressed 5G as the aorta and AI as the brain.

▲ 5G is expected to drive increased demand for semiconductors in autonomous vehicles, AI robots, and Internet of Things devices.
5G, characterized by ultra-high speed, hyper-connectivity, and ultra-low latency, maximizes edge computing performance and helps AI analysis that requires real-time large-scale data processing. It minimizes the process of sending and receiving data in a cloud environment, thereby reducing the burden on the cloud and increasing data processing speed, while securing competitiveness in terms of security, data storage, and processing costs.
In particular, in the era of the 4th industrial revolution, 5G is driving the semiconductor market. Experts predict that demand for semiconductors will surge in new areas such as self-driving cars, artificial intelligence robots, and Internet of Things devices.
Samsung Electronics and SK Hynix are the world's strongest in the global semiconductor market, but the situation is not one of complacency. It is because of China, which is catching up at a fierce pace, focusing on 5G and AI. In the era of the 4th Industrial Revolution, from Made in China to Made for China, we learned about the future of the semiconductor industry from SK Securities IT General Manager Kim Young-woo.
China challenges, America blocks Huawei participated in the world's largest mobile trade show, 'MWC 2019' held in February and put forward the slogan 'Building a Fully Connected, Intelligent World', and is promoting a long-term strategy of using 5G for artificial intelligence to produce and distribute an AI computing platform.

▲ China was planning to push for more than half of its telecommunications equipment to be from Huawei, but the US-China conflict has become an obstacle.
5G has the characteristic of using a lot of data, so if 5G technology develops, it can be applied beyond mobile phones to the AI field. Huawei is significant in that it is a total solution company that possesses the technology for 5G, AI, and even RPA (Robotic Process Automation) that applies and executes AI.
Currently, Europe has the highest level of RPA, home to companies such as SAP and Siemens, and the US has the best in machine learning and cloud technology, but experts predict that China will take the lead in the next two to three years.
Chinese companies promise to provide personal information whenever the Chinese government requests it, and the amount of sample data that forms the basis of AI is overwhelming. The core of AI technology depends on how much the technology can be put to practical use, and if the initial data collection is well-equipped, AI technology can develop at a rapid pace.
Accordingly, the United States is using its strengths in semiconductors and OS to attack Huawei in order to prevent it from giving up its leadership in the global market, and the British company Arm is joining in.
The US-China trade dispute has evolved beyond an economic war into a war for hegemony.
5G·Edge Computing Mutually Complementary Relationship Recently, the number of cases of applying IoT technology in various industrial fields is increasing. However, the infrastructure for sending data generated from IoT to the server is not yet properly established. Data generated from IoT must be sent to the cloud server one by one, and security issues may arise here.
Experts point out that with the introduction of 5G, we must develop edge computing technology that processes data on the terminal itself or around the terminal. Above all, edge computing technology can play a role in filtering out unwanted information (data), so it is expected to be applied to various fields beyond the communications field, such as artificial intelligence speakers and self-driving cars.
As most edge computing development companies in Korea have been fabless companies, commercialization of edge computing technology could have a positive effect on foundry companies as well.
China aims to become a global powerhouse with 5G technology Currently, due to the aftermath of the US-China conflict, HiSilicon, the number one company in the Chinese system semiconductor industry and a subsidiary of Huawei, is under sanctions, but UNISOC, the number two company in the industry, is attracting attention in the global market by recently releasing a 5G model.

▲ China is expected to use the 2022 Beijing Winter Olympics as an opportunity to accelerate its 5G strategy.
President Xi Jinping had planned to release a 5G phone priced at just $600 in the second half of 2019 and a 5G phone priced at $300 in 2020 through HiSilicon ahead of the 2022 Beijing Winter Olympics, but the plan fell through due to the US-China conflict.
However, since 2022 is also significant as the year that President Xi Jinping marks the 10th anniversary of the National People's Congress, it is expected that he will pursue Plan B to showcase China's 5G technology through the Olympics.
According to China's 'Made in China 2025' announced in 2015, China will become the world's leading manufacturer of integrated circuits (While it showed weaknesses in semiconductors, integrated circuit design (semiconductor manufacturing equipment), and civil aviation, it showed strengths in telecommunications, railways, and solar energy businesses.
Based on this, China has been promoting the 'One Belt, One Road' project from 2014 to 2049, which will create synergy effects in telecommunications, railways, and solar energy businesses. This large-scale project, which aims to build a modern-day Silk Road, a transportation route between the East and the West, and open the way for expanded economic and trade cooperation between China and neighboring countries, has, however, been put on hold due to U.S. sanctions.
Accordingly, China has shown a strategy of focusing on securing fabless design technology in the semiconductor sector and then seeking the next step. It is noteworthy that China’s memory sector was not affected by the US attack. Considering that Samsung Electronics and SK Hynix had high memory supply in 2018, experts say that this actually worked to China’s advantage. For domestic companies, it also served as an opportunity to advance in the system semiconductor foundry sector, as they could supply Micron, an American company, through Samsung Electronics and Hynix.
Korea continued production because it was in a gray zone (neutral country) with no reason to stop semiconductor production in the US-China conflict, which ultimately worked to China's advantage.
Is Arm's Huawei Attack Because of the Vision Fund? Looking at the shareholder composition of the British company Arm, 75% is SoftBank, and 25% is SoftBank's Vision Fund. Major customers include Samsung Electronics and Huawei, which use all graphics chips up to Mali, as well as Apple and Qualcomm, which use their own graphics processing unit Adreno but receive some architecture.
Even considering only the cost aspect, Huawei must be paying a high cost, so why did Arm take the position of an attacker against Huawei?
Experts are speculating that SoftBank's Vision Fund is in bad shape, so it attacked Huawei as a breakthrough. In the aftermath of the murder of Jamal Khashoggi in October 2018, SoftBank's stock price began to fall, and since then, WeWork, which is part of the Vision Fund, has also seen its stock price plummet by 50%, and Uber's stock price has fallen since its first day of listing.
However, from the perspective of Softbank, which needs to complete Vision Fund 2, Kim's view is that it targeted Saudi Arabia and the United States. In the United States, there are four telecommunications companies: Verizon, AT&T, Sprint, and T-Mobile, and they are famous for not competing with each other. In addition, the recent acquisition of Sprint by T-Mobile (Softbank owns an 84% stake in Sprint) can be seen as meaning that they will eventually apply 5G technology to 99% of the entire United States, including mobile phones and artificial intelligence. In fact, based on these conditions, they received approval from the U.S. Department of Justice in July and plan to actively utilize 5G and AI technologies starting in 2020.
In the Middle East, the first Vision Fund was successfully funded thanks to Saudi Aramco's 'NEOM MEGA CITY PROJECT', but the funding of the second Vision Fund is currently uncertain, and even the first Vision Fund is said to be in danger.
Securing semiconductor process technology is the key It is unclear whether Intel will be able to mass produce 7nm semiconductors due to limitations in its current process technology. With Intel's CPU sales in crisis, Intel is marketing with the concept of cost-effectiveness.
Huawei uses Cortex A 76 due to US sanctions, while Samsung uses Cortex A 77. However, ironically, Huawei is ahead of Samsung when it comes to chip speed. This is due to the difference in process technology, with Samsung’s chip being 8nm while Huawei’s is equipped with a 7nm chip.
The current state of semiconductor technology in the automobile market is also Arm-based. Qualcomm has its semiconductors installed in all vehicles, but Samsung Electronics has them only in the Audi A4. This means that improving process technology is an urgent task.

▲ Samsung Electronics' Galaxy S 11, expected to feature Arm Cortex A77
Samsung Electronics leaps to No. 1 in non-memory semiconductors with 3GAE
Samsung Electronics distributed the 3GAE (3nm Gate-All Ariund Early) process design kit through the 'Samsung Foundry Forum 2019' held in May. When the line width is reduced to 3nm, the GAA structure with gates on all four sides is selected.
This is a 45% reduction in area compared to the latest 7nm-based chip, reducing power consumption by 50% while increasing performance by 35%. Samsung Electronics has expressed its intention to share its 3nm-based process technology with fabless companies to enable more companies to use GAA-based MBCFET technology in the future.
The typical GAA channel region requires high operating voltage and has difficulty in flowing a lot of current because it uses a thin and long nanowire shape. Samsung Electronics has implemented a technology to stack channels using its own Multi Bridge Channel FET (MBCFETTM) technology that uses nanosheet shapes in the channel, thereby reducing the operating voltage by 50%.
In order to rise to the top non-memory semiconductor company in 2030, it is necessary to increase the market share of Foundry and System LSI. Samsung Electronics plans to start mass production of 5nm in 2020 and mass production of 3nm GAE in 2021.
TSMC, the world's number one foundry, is negative about this because it is impossible to implement nanosheets in the 2nm process, and argues that material development should take priority in order to use the nanosheet structure.
Experts advise that the future of corporate process technology will depend on how to secure the reliability and durability of GAA in the 2nm process.