ARM CSS Partner

Silicon for
the AI Data Center

Anovus Labs designs next-generation server computing chips built on ARM Compute Sub-Systems. Malaysian engineering, global ambition — powering hyperscale AI workloads with unprecedented efficiency.

✦ Made in Malaysia · Powered by ARM

Explore Our Technology

Why Anovus Labs

Building the foundation for Malaysia's semiconductor sovereignty

🧬

ARM CSS Architecture

Purpose-built on ARM's Compute Sub-System platform — the same foundation trusted by hyperscalers worldwide for custom silicon.

⚡

Performance per Watt

Data centers demand efficiency. Our designs deliver >2× performance-per-watt over legacy architectures, slashing TCO.

🛡️

Security by Design

Integrated TrustZone, hardware root of trust, and confidential compute capabilities — security isn't bolted on, it's baked in.

🧱

Chiplet-Ready

Modular chiplet architecture with die-to-die interconnects. Scale from edge to exascale without redesigning from scratch.

🇲🇾

Malaysian Innovation

Part of Malaysia's Silicon Vision — building indigenous IP and a complete semiconductor ecosystem on home soil.

🤝

Ecosystem Partner

Collaborating with foundries, EDA vendors, and system integrators to deliver turnkey silicon — from concept to production.

ARM CSS Inside

The Compute Sub-System that powers our silicon

ARM Neoverse N2

Armv9 infrastructure CPU — 40% IPC uplift over N1. Built for cloud-to-edge scalability. Pre-validated platform with integrated memory controllers, PCIe Gen5/CXL, and multi-die coherent interconnects.

Neoverse N2 Cores

Up to 64 Armv9 cores per CSS — 40% IPC uplift over N1 with market-leading performance-per-watt

DDR5 & PCIe Gen5/CXL

Integrated high-bandwidth memory and I/O with CXL-attached device support for scalable designs

Pre-Validated Platform

CPU, memory, and I/O engineered as a single verified architecture — concept to silicon, faster

Multi-Die Ready

Combine multiple CSS instances with coherent interconnects for socket-level scalability beyond 64 cores

Solutions

Chip designs for the workloads that matter

AI Inference Server

High-throughput transformer inference leveraging Neoverse N2's SVE2 vector extensions. Optimized for cloud-scale AI serving.

40%
IPC Uplift vs Neoverse N1

Cloud-Native DPU

Data Processing Units with integrated PCIe Gen5 and CXL for networking, storage, and security offload.

5nm
Advanced Process Node

General Compute SoC

Pre-validated ARM CSS platform with up to 64 N2 cores per subsystem. Multi-die scaling for hyperscale deployments.

64
N2 Cores per CSS