TSMC and ASIC business model for AI ASIC development unsung heroes

Artificial intelligence-related special applications of integrable circuits (ASIC) Recently, the market has been gradually noticed that a number of industries such as NVIDIA, Intel, Google and some of the new ventures have been in the development, is expected to form the future of the billions of dollar market opportunities, as far as these ASIC chip design behind, As the biggest thrust to provide support for the advent of AI chips, the ASIC business model and TSMC are in fact. According to the Semiwiki Web site, as Nvidia repositioned the graphics chip (GPU) into the role of the cloud AI engine, it is also determined to drive the ASIC business forward, such as Google now has a tensor processor (TPU), Intel after buying Nervana company also made its own Nervana chip, in addition by a number of former Google TPU staff founder of the new Enterprise GROQ, recently announced that it will be launched in 2018, the next generation of AI chip, and realize these AI chip dream behind the hand, Non-TSMC and ASIC business model. This from the United States amorphous wafer factory FinFET-level ASIC design service provider Esilicon recently announced that the success of its own deep learning (DL) ASIC sent to the manufacture of the release of the release of the clues. Esilicon referred to this ASIC production of IP, advanced 2.5D packaging process as well as one of the industry's largest chips, and for the company's first use of the TSMC 2.5D COWOS packaging technology production chips. Mr. BJ Woo, vice president of TSMC's business development, said that the Cowos packaging technology is designed to meet the needs of this type of chip design depth learning application, this advanced package solution can achieve high-performance and integration requirements to achieve Esilicon design goals. According to the report, all of the chips mentioned above were manufactured by the ASIC and referred to TSMC, in which case it would be worthwhile to see if Intel would manufacture Nervana chips in its own production line, as Nervana is now bought by Intel, but outside media analysis, It is easier said than done to move the chip manufacturing from the TSMC side to the Intel side. Nervana was purchased by Intel, and its CEO Naveen Rao believes that the way the chip is designed to go beyond Nvidia is not toward the computer gaming field, but rather for the neural network, the Nervana business must be integrated into Intel's other business, and the AI chip will not run alone, A period of time will be associated with Intel's still-leading cloud data Center central processing Unit (CPU) in the world. GROQ, a secretive new venture of 8 co-founders from Google's initial 10-person team, has been preparing to share its first AI chip with the outside world in 2018, although it has only received $10 million to raise capital. From above, it shows that the current ASIC business model is developing properly, so that more and more AI ASIC can be introduced and developed, and the implementation of the finished product is based on the TSMC process.

2016 GoodChinaBrand | ICP: 12011751 | China Exports