2026-06-01 Facebook Twitter LinkedIn Axelera AI has integrated the AndesCore™ AX65 Out-of-Order RISC-V processor into its newly unveiled Europa AI Processing Unit (AIPU) to provide a powerful, open-standard foundation for next-generation computer…
Facebook
Twitter
LinkedIn

Axelera AI has integrated the AndesCore™ AX65 Out-of-Order RISC-V processor into its newly unveiled Europa AI Processing Unit (AIPU) to provide a powerful, open-standard foundation for next-generation computer vision and generative AI at the edge.
EINDHOVEN, Netherlands & HSINCHU, Taiwan — June 1, 2026 — Axelera AI, the leading provider of high-performance, ultra-efficient edge AI solutions, and Andes Technology (TWSE: 6533), a premier supplier of high-efficiency 32/64-bit RISC-V processor cores and a Founding Premier Member of RISC-V International, today announced a strategic partnership. Axelera AI has integrated the AndesCore™ AX65 Out-of-Order RISC-V processor into its newly unveiled Europa AI Processing Unit (AIPU) to provide a powerful, open-standard foundation for next-generation computer vision and generative AI at the edge.
As AI models—particularly Large Language Models (LLMs) and high-resolution vision transformers—move from data centers to edge environments, the need for a balanced architecture of high-performance control and specialized AI acceleration has reached a critical point. The Europa platform addresses this by combining Axelera’s proprietary Digital In-Memory Compute (D-IMC) technology with the sophisticated control capabilities of the Andes AX65. With 629 TOPS and integrated pre- and post-processing, Europa extends AI capabilities to edge servers and unlocks AI applications that more compact devices can’t handle. It consolidates workloads across facilities, deploys robotics and autonomous systems, and runs complex models locally without datacenter infrastructure.
**The Synergy: D-IMC Meets Out-of-Order RISC-V
**The Europa AIPU delivers a staggering 629 TOPS of peak performance. While Axelera’s D-IMC cores handle 99% of the heavy lifting for neural network inference, the AndesCore™ AX65 serves as the primary “brain” of the chip.
Key benefits of this integration include:
- Advanced Control Flow: The AX65’s 13-stage, 4-wide out-of-order execution can handle multiple complex pre- and post-processing tasks in parallel, such as frame management and result formatting, with significantly lower latency than traditional in-order cores.
- Linux-Ready Performance: Compliant with the RVA22 profile, the AX65 provides the robust architectural features required to run full Linux distributions, simplifying software deployment for developers.
- Power Efficiency: By leveraging the AX65’s superior performance-per-watt, Europa maintains its market-leading efficiency, delivering up to 3x the performance-per-watt of legacy GPU-based solutions.
“The integration of the Andes AX65 into our Europa architecture marks a significant milestone in our mission to democratize AI,” said Giuseppe Garcea, Director of Silicon & Co-founder of Axelera AI. “By pairing our disruptive D-IMC technology with Andes’ world-class RISC-V expertise, we are providing our customers with a platform that is not only incredibly fast but also highly flexible and easy to program.”
“We are thrilled to see our high-performance AX65 core at the heart of Axelera AI’s groundbreaking Europa processor,” said Dr. Charlie Su, President and CTO of Andes Technology. “The AX65 was designed to meet the rigorous demands of infrastructure and high-end AI SoCs. This partnership demonstrates how the RISC-V ecosystem is now capable of competing at the highest levels of performance, enabling a new era of edge intelligence.”
**Availability
**The Europa AIPU, featuring the AndesCore™ AX65, is currently sampling to lead customers.
**
About Axelera AI
**Axelera AI is delivering the world’s most powerful and advanced solutions for AI at the edge. Its game-changing hardware and software products integrate into a variety of edge devices, from small industrial modules to high-density servers, enabling the most advanced AI applications. Headquartered in Eindhoven, The Netherlands, Axelera AI has a global team of experts in AI, semiconductors, and software.
**About Andes Technology
**As a Founding Premier member of RISC-V International and a leader in commercial CPU IP, Andes Technology (TWSE: 6533
; SIN: US03420C2089
; ISIN: US03420C1099
) is driving the global adoption of RISC-V. Andes’ extensive RISC-V Processor IP portfolio spans from ultra-efficient 32-bit CPUs to high-performance 64-bit Out-of-Order multiprocessor coherent clusters. With advanced vector processing, DSP capabilities, the powerful Andes Automated Custom Extension (ACE) framework, end-to-end AI hardware/software stack, ISO 26262 certification with full compliance, and a robust software ecosystem, Andes unlocks the full potential of RISC-V, empowering customers to accelerate innovation across AI, automotive, communications, consumer electronics, data centers, and mobile devices. Over 20 billion Andes-powered SoCs are driving innovations globally. Discover more at www.andestech.com
and connect with Andes on LinkedIn
, X (formerly Twitter)
, YouTube
and Bilibili
.
The collaboration expands production-grade virtualization across Andes’ processor portfolio, starting with immediate availability of Bao Hypervisor on AndesCore™ AX66.
Guimarães, PT, May 11, 2026 — OSYX Technologies, the company behind the open-source Bao Hypervisor, today announced a strategic collaboration with Andes Technology Corporation, a leading provider of high-performance, low-power 32- and 64-bit RISC-V processor IP and a Founding Premier Member of RISC-V International, to support virtualization across Andes’ RISC-V processor portfolio. As an initial milestone, Bao Hypervisor is now available on the AndesCore™ AX66, a high-performance application-class processor.
AndesCore™ AX66 is an out-of-order, 13-stage pipeline, 4-wide decode, and 8-wide out-of-order execution multicore processor IP supporting the RVA23 profile, Vector and Vector Crypto support, RISC-V Hypervisor Extension (H-extension) and AIA. Each core has 64KB private L1 instruction and data caches and up to 1MB private L2 cache, and each cluster contains up to 8 cores and a shared L3 cache up to 32MB.
Bao delivers a static partitioning virtualization model that enables multiple software domains to run concurrently on a single platform, including rich operating systems and real-time workloads, with strong isolation and predictable performance. On AX66-based systems, Bao supports key system-level IPs out of the box, including the Advanced Interrupt Architecture (AIA) and Input–Output Memory Management Unit (IOMMU), enabling efficient interrupt virtualization, device isolation, and scalable integration for complex SoCs.
“Virtualization is becoming a key building block for next-generation compute platforms,” noted Sandro Pinto, Co-founder of OSYX Technologies and active contributor within RISC-V International. “Supporting Bao on Andes’ AX66 processors brings production-grade virtualization to a widely deployed RISC-V platform and reinforces a consistent approach to system architecture.”
“Virtualization is critical for modern SoC design, and we are excited to partner with OSYX Technologies to bring the Bao Hypervisor to our AndesCore™ AX66.” said Dr. Charlie Su, President and CTO of Andes Technology. “By combining our RVA23-compliant hardware with Bao’s static partitioning, we provide customers with a high-performance, secure, and production-ready platform that simplifies the integration of complex real-time and rich OS workloads.”

The collaboration establishes a path for broader support across Andes processors with virtualization capabilities, with additional platforms expected to follow, while laying the groundwork for continued alignment in system-level software and platform security.
Target platforms include high-performance embedded and edge systems aligned with AX66 deployments, such as industrial automation and robotics platforms, automotive and mobility subsystems, networking and infrastructure equipment, and intelligent edge devices combining application processing, AI workloads, and real-time control. By enabling efficient integration of diverse software domains on a single platform, Bao supports scalable system architectures aligned with next-generation RISC-V platforms.
Both organizations are active contributors to the RISC-V ecosystem, participating in technical working groups and shaping the evolution of ISA and system-level specifications. This collaboration reflects the continued maturation of RISC-V toward production-ready virtualization across a broad range of platforms.