Alibaba Unveils RISC-V Chip in Push for Chinese Semiconductor Independence
24 March 2025 路 Uncategorized 路
Source: 路 https://technews.tw/2025/03/14/alibaba-risc-v-china-tech/
The Alibaba Group's server-level XuanTie C930 processor is poised to accelerate the adoption of open-source chips based on the RISC-V instruction set architecture in China, according to analysts. This move also aims to counter US technological restrictions.
Alibaba鈥檚 DAMO Academy announced that shipments of the C930 will begin this month but did not disclose specific figures. The CPU design can be licensed by integrated circuit (IC) developers for use. Academician Niu Guangnan from the Chinese Engineering Academy stated at a press conference, "The open-source model will help establish an inclusive and collaborative global RISC-V ecosystem to become the new engine disrupting the semiconductor industry." The South China Morning Post believes that Alibaba's latest initiative demonstrates significant progress in its RISC-V chip design over recent years and reflects China鈥檚 determination to overcome US sanctions.
Analysts at Guotai Junan Securities suggest, mirroring DeepSeek鈥檚 disruption of OpenAI with an open source model offering low cost and high performance, the potential for RISC-V is substantial within the AI era.
DeepSeek achieved breakthroughs through its advanced open-source AI models V3 and R1, requiring significantly less computing power and expense compared to those needed by large tech companies developing Large Language Models (LLMs). Open source allows public access to program code, enabling third-party developers to modify or share designs, fix issues, or expand functionalities. This approach has already made significant contributions to China's technological industry over the past few decades.
Beijing semiconductor and computing system expert Li Yangwei believes that opportunities will increase in the coming years following recent efforts to establish a unified and stable software ecosystem for RISC-V. For Chinese chip designers, RISC-V now presents an alternative solution to proprietary architectures like Intel鈥檚 X86 architecture dominating PCs or Arm's monopoly on mobile devices.
At last month's launch event, DAMO Academy stated that its IC design team has supported "over 30% of high-performance processors based on the RISC-V instruction set." Previously, it released several processors utilizing this framework, including the XuanTie C920 (last year) and the XuanTie C910 in 2019. Furthermore, DAMO Academy announced plans to develop multiple new Xuanties: the C908X for AI acceleration, R908A for automotive applications, and XL200 for high-speed interconnect design.
Prior to Alibaba's increased focus on RISC-V, chips using this architecture were primarily applied in relatively low-end devices like headphones; their application in laptops or server processors remained limited. China is expected to further support the development of RISC-V with Reuters reporting that a guideline encouraging nationwide adoption could be released this month.
According to a report from market research firm SHD Group, global shipments of RISC-V system-on-chips (SoCs) are forecasted to increase from 13 billion units in 2023 to 16.2 billion by 2030, with the market size reaching $92.7 billion.
(Lead image source: Shutterstock)
Alibaba鈥檚 DAMO Academy announced that shipments of the C930 will begin this month but did not disclose specific figures. The CPU design can be licensed by integrated circuit (IC) developers for use. Academician Niu Guangnan from the Chinese Engineering Academy stated at a press conference, "The open-source model will help establish an inclusive and collaborative global RISC-V ecosystem to become the new engine disrupting the semiconductor industry." The South China Morning Post believes that Alibaba's latest initiative demonstrates significant progress in its RISC-V chip design over recent years and reflects China鈥檚 determination to overcome US sanctions.
Analysts at Guotai Junan Securities suggest, mirroring DeepSeek鈥檚 disruption of OpenAI with an open source model offering low cost and high performance, the potential for RISC-V is substantial within the AI era.
DeepSeek achieved breakthroughs through its advanced open-source AI models V3 and R1, requiring significantly less computing power and expense compared to those needed by large tech companies developing Large Language Models (LLMs). Open source allows public access to program code, enabling third-party developers to modify or share designs, fix issues, or expand functionalities. This approach has already made significant contributions to China's technological industry over the past few decades.
Beijing semiconductor and computing system expert Li Yangwei believes that opportunities will increase in the coming years following recent efforts to establish a unified and stable software ecosystem for RISC-V. For Chinese chip designers, RISC-V now presents an alternative solution to proprietary architectures like Intel鈥檚 X86 architecture dominating PCs or Arm's monopoly on mobile devices.
At last month's launch event, DAMO Academy stated that its IC design team has supported "over 30% of high-performance processors based on the RISC-V instruction set." Previously, it released several processors utilizing this framework, including the XuanTie C920 (last year) and the XuanTie C910 in 2019. Furthermore, DAMO Academy announced plans to develop multiple new Xuanties: the C908X for AI acceleration, R908A for automotive applications, and XL200 for high-speed interconnect design.
Prior to Alibaba's increased focus on RISC-V, chips using this architecture were primarily applied in relatively low-end devices like headphones; their application in laptops or server processors remained limited. China is expected to further support the development of RISC-V with Reuters reporting that a guideline encouraging nationwide adoption could be released this month.
According to a report from market research firm SHD Group, global shipments of RISC-V system-on-chips (SoCs) are forecasted to increase from 13 billion units in 2023 to 16.2 billion by 2030, with the market size reaching $92.7 billion.
(Lead image source: Shutterstock)