阿里达摩院玄铁RISC-V IP累计出货突破60亿颗并发布车规E908A，但属生态里程碑、无订单与收入增量

Cumulative shipments of chips incorporating Alibaba DAMO Academy's RISC-V processor intellectual property (IP), branded XuanTie, have surpassed 6 billion, marking a milestone for the open-source architecture as the company expands beyond embedded applications into AI computing, servers, autonomous driving, and robotics.

Dr. Jianyi Meng, chief scientist at DAMO Academy, disclosed the shipment figure at the sixth RISC-V China Summit, held in Shenzhen from October 18 to 20, 2026, according to C114.net and STAR Market Daily . The company also introduced the XuanTie E908A, its first automotive safety-enhanced processor, extending its portfolio into safety-critical automotive applications.
The milestone reflects the growing commercial adoption of RISC-V, an open instruction set architecture that allows chip designers to develop customized processors. According to Omdia, global RISC-V processor shipments are projected to grow by nearly 30% annually through 2030, when the architecture is expected to account for approximately 25% of worldwide processor shipments.
From embedded chips to high-performance AI servers
XuanTie's cumulative shipments have increased from more than 4 billion in 2024 to over 6 billion. Its processor IP spans the C series for high-performance computing, the R series for reliable real-time applications, and the E series for low-power embedded systems, alongside XT-Link multicore interconnect technology.
The IP is already deployed across microcontrollers, wireless connectivity, display controllers, audio-video processing, and security applications. Alibaba DAMO Academy is now seeking to extend that commercial base into higher-performance computing.
The company's flagship XuanTie C950, introduced in March 2026, is based on the RV23 standard and achieved a single-core SPECint2006 benchmark score exceeding 70, targeting cloud computing, generative AI, advanced robotics, and edge computing.
Meng said partners are advancing XuanTie-based products for server baseboard management controllers, networking, GPU storage controllers, graphics processors, and accelerators, with server products incorporating the latest C950 expected to emerge in 2027.
The expansion reflects changing CPU requirements in AI infrastructure. Meng noted that agentic AI is placing greater demands on CPUs to handle complex workloads with higher performance, lower latency, and manageable costs.
To address these requirements, DAMO Academy is developing XuanTie as an architecture capable of supporting AI computation natively rather than relying solely on external accelerators.
The company has integrated scalar, vector, and matrix operations alongside multilevel memory architecture and specialized acceleration for Transformer operations, as EET China noted. It reported operator utilization above 80%, memory bandwidth utilization exceeding 90%, and performance improvements of two to ten times for selected AI operations.
DAMO Academy also reported that a 128-core C950 configuration maintained approximately 78% efficiency, while its multicore system outperformed a competing x86 platform by around 11% across four agentic AI workloads.

DAMO Academy chief scientist Dr Jianyi Meng. Credit: DAMO
Automotive processors extend RISC-V's commercial reach
Automotive computing represents another priority. The newly unveiled XuanTie E908A, derived from the E908 platform, is designed for applications including automotive safety islands, domain controllers, and in-vehicle gateways.
DAMO Academy said the processor delivers 20% higher performance than comparable current-generation products and achieves a single-point fault metric (SPFM) exceeding 90%. The company also described the E908A as its first processor designed to address Automotive Safety Integrity Level (ASIL) B and D requirements.
The E908A is part of a broader automotive and embodied AI roadmap that includes the planned C925A, C930A, and C950A processors. DAMO Academy expects the C950A to deliver five times the overall performance of its first-generation automotive processor.
The roadmap reflects a shift towards applications where real-time responsiveness, functional safety, and computing performance must be developed together, extending RISC-V's relevance beyond its established embedded-device market.
Software compatibility remains key to RISC-V adoption
Hardware advances alone cannot secure RISC-V adoption in enterprise and AI infrastructure, according to Meng, who identified software compatibility and ecosystem fragmentation as key challenges. While RISC-V's flexibility enables customized architectures, differences between platforms complicate software deployment.
DAMO Academy is addressing these gaps through compiler, Java runtime, and operating system development. It has contributed 43 patches to the GNU Compiler Collection (GCC), accounting for 17.5% of relevant contributions, second only to Qualcomm, and delivering approximately 10% performance gains in typical workloads.
Its XuanTie team has also submitted 111 LLVM patches and, over six months, contributed 48 OpenJDK patches that improved the C950's SPECjbb performance by 31%. Collaboration with the Dragonwell team has further produced a Java runtime for Alibaba's e-commerce operations.
The company is also working with Ubuntu, Alibaba Cloud, and NeoKylin on commercial operating system support, while contributing to the RISC-V Software Ecosystem (RISE) initiative through Android Open Source Project compatibility and automated testing.
Meng described RISC-V's fragmentation as transitional, with greater standardization and ecosystem coordination expected to improve software compatibility.
Meanwhile, DAMO Academy is exploring simultaneous multithreading with Zhejiang Lab and ZTE Microelectronics. Initial two-thread testing on the C950 delivered a 35% performance improvement, while the company targets aggregate single-core multithreaded SPECint2026 performance exceeding 100 points by 2028.
"To bring high-performance RISC-V into mass production, benchmark processor products alone are not enough," Meng said, emphasizing continued hardware and software development alongside engineering validation.
For DAMO Academy, the priority is translating XuanTie's shipment scale and technical advances into production-ready products across AI computing, automotive electronics, and intelligent machines.