Welcome to Hantech Electronics Ltd. | Register

Home > Industry Information > World Core Business Until The End Of The First 7 Nanometer Asic Years To Complete

World Core Business Until The End Of The First 7 Nanometer Asic Years To Complete

Auth: Date:2018/5/22 Source:laoyaoba.com Visit:6507 Related Key Words: World core 7 nanometers ASIC

The application of artificial intelligence (AI) has been widely used in many fields, such as cloud computing, high efficiency computing (HPC), self driving, encrypting money and block chain, intelligent mobile phone edge operation. In order to differentiate the industry from the same industry, the operation processor has turned from GPU to the development of the special application chip. (ASIC).


IC design service plant world core -KY (3661) is benefited from the AI operation ASIC's client design (NRE) and the mass production order influx. The operation is prosperous to the end of the year, and the first 7 nanometers ASIC is expected to complete the design decision by the end of this year.


AI applications spread rapidly to industries in the second half of last year, most of which were in the early stages, but encrypted currency and block chain ASIC took the lead in becoming the focus of the market because the mining machines like the bit continent wanted to make differentiated products and use ASIC to improve computing efficiency. Since the start of mining ASIC, such as bitcoin, is very successful, including apple, Samsung, Google, Amazon, Alibaba, HUAWEI and other international system major factories have decided to develop AI operation ASIC, and most of the orders are made by creative, -KY and other IC design service providers.


In fact, at the present stage, all the system plants are using GPU to perform operation, but the current AI operation types are multiple and complex, and there is no common AI operation mechanism. The calculation methods that different industries or different system factories want to adopt are different. If we continue to use GPU as a AI computing processor, it will eventually become an operation. The results are all the same, so it is difficult to differentiate. Therefore, in order to carry out AI operation for the useful data of the system, the system factory has a high demand for designing the customized AI operation ASIC.


For the system plant, developing a customized AI operation ASIC is a trend, but most of the system plants themselves have no IC design team, and the AI operation should adopt 16 nanometers or 7 nanometers and other advanced processes. The cost of developing 16 nanometer AI operation ASIC is up to $100 million, and the development of 7 nanometer AI operation ASIC is more than $3~4 billion. Therefore, in order to reduce the risk of designing ASIC, outsourcing to experienced IC design service factories is the best solution.


-KY has worked with Nihon University to develop the experience of supercomputer computing ASIC in the past, so the AI operation ASIC market began to be laid out last year. It was initially locked in the HPC operation related applications, has already obtained the Japanese and mainland customer orders, and has completed the ASIC design finalize. It is expected that the second and three quarter will begin to enter the scale production order. Paragraph, customer demand is a million take-off, revenue will be obvious note. In addition, the first 7 nanometer AI operation ASIC of -KY will also be finalized by the end of the year. Next year's mass production will bring greater revenue contribution.


-KY first quarter combined revenue season increased by 11.2% to 812 million yuan, net profit after tax 113 million yuan, compared with the same period last year, a significant growth of more than 2 times, net profit of 1.85 yuan per share, better than the market expectations. World core -KY announcement April consolidated revenue 333 million yuan, legal person optimistic about second quarter revenue will see 1 billion yuan. -KY does not comment on the financial figures of the legal person.

Industry Information

Product Index :