Alibaba’s R&D arm DAMO Academy announced yesterday that it is developing a new neural network chip called Ali-NPU for AI inferencing in the field of image processing, machine learning, etc.
Ali-NPU is expected to be some 40 times more cost-effective than conventional chips. Alibaba Researcher Yang Jiao says this chip’s performance will be 10 times better than mainstream CPU and GPU architecture AI chips currently on the market, with only half the manufacturing cost and power consumption. By self-developing AI chips tailored to its own needs, Alibaba will reduce its dependence on other chip companies.
In 2015, Alibaba cooperated with Hangzhou C-SKY Microsystems to develop the Yun On Chip (YoC). In November 2016, Alibaba and Tencent invested US$23 million in California chip company Barefoot Networks.
DAMO Academy is an R&D institute for fundamental and disruptive technology research established by Alibaba in November 2017. It has a dozen top-notch international scientists working in quantum computing, machine learning, network security, visual computing, natural language processing, chip technology, and embedded technology. Alibaba is planning to inject US$15 billion into this project over the next three years.
An update on the story:
Alibaba announced yesterday that it has acquired chipmaker C-Sky Microsystems in a bid to strengthen its chip-making R&D capability. Financial terms of the deal were not disclosed.
Founded in 2011, Hangzhou-based C-Sky Microsystems is an integrated circuit design company dedicated to 32-bit, high-performance and low-power embedded CPU, with chip architecture licensing as its core business. The company’s embedded CK-CPU chip software and hardware platform provides customers with CPU IP core, System-on-Chips (SoC) design and development platforms.
C-Sky Microsystems bills itself as “the only embedded CPU volume provider in China with its own instruction set architecture.”
Author: Paul Fan | Editor: Tony Peng, Michael Sarazen