The move is being made to boost Alibaba's capabilities in datacenters for the Internet and particularly for the Internet of Things as C-Sky has focused on 32-bit embedded processor design. Alibaba was previously an investor in C-Sky but has now converted that minority investment into almost total ownership.
C-Sky was founded in 2001 and has developed a proprietary series of 32-bit C-SKY embedded CPU cores and architecture aiming at ultra-low power, high performance, high code density and ease of use. The company claims to be the only embedded CPU volume provider in China with its own instruction set architecture. Approximately 600 million chips have shipped with C-Sky embedded CPUs, according to report, but mostly in China.
Alibaba has previously invested in five chip manufacturers, including AI chip designer firms Cambricon and Kneron and networking chip developer Barefoot Networks (see Qualcomm-backed startup announces AI processor family).
Meanwhile, the company is reportedly developing its own artificial neural network acceleration processor, the Ali-NPU, for use in artificial intelligence applications such as image video analysis, machine learning.
Bringing proven processor design capability in-house could certainly help with that.
It is notable that C-Sky is a top-tier, platinum member of the RISC-V Foundation suggesting the company is not wedded to its own proprietary architecture for the future and will perhaps use the modularity of RISC-V to develop 64-bit processor cores or to add machine learning capabilities.
With the acquisition of C-Sky, Alibaba is following a trend that can be traced back almost exactly eight years to Apple's acquisition of PA Semiconductor for $278 million. Prior to that time the industry had gone through a process of disaggregation, separating system companies from the IC manufacturing and semiconductor design that underpinned their systems.
Since that time the larger product and service companies – such as Apple. Google, Amazon, Facebook have been re-engaging with processor design, but not with the capital-intensive manufacturing, to create chips that are suited to their particular requirements.
Related links and articles: