CSS Overview

What are the Arm Compute Subsystems (CSS)? 

Arm Compute Subsystems (CSS) are pre-integrated, market-tuned subsystems that combine optimized compute, interconnects and system IP. Arm CSS isn’t a single, static product; it’s a modern way to deliver the Arm technology portfolio, purpose-built for key markets like mobile, infrastructure, and automotive. They help partners shorten design cycles, accelerate time to market, and achieve top-tier performance. From the cloud to the edge, CSS provides the speed, flexibility, and AI-ready foundation needed to build the next generation of innovation.

Why it Matters

Why Arm Compute Subsystems Matter?

CSS helps partners overcome design complexity by enabling a platform-first approach that cuts integration time and cost. Instead of assembling cores, memory, security, and interconnect from scratch, partners begin with a validated compute foundation tailored to their target application.


Benefits of Arm CSS:

arm icon accelerate development

Accelerates Development

Partners report up to 12 months faster time to market.

arm icon reduce cost and effor

Reduces Cost and Effort

Tens of millions in non-recurring engineering (NRE) savings.

arm icon chip

Optimized for AI

Future-ready compute in every market.

arm icon customizable foundation

Customizable Foundation

Built to support differentiated design.

arm icon market scalability

Cross-market Scalability

From cloud to edge.

What Makes Up Arm Compute Subsystems? 

Arm Compute Subsystems include the essential compute components needed to build high-performance, scalable SoCs, pre-integrated to reduce design complexity and accelerate delivery. While each implementation is specific to the market use case, each CSS typically contains:

Arm CPUs

Latest-generation cores tailored for each market (e.g., Arm Neoverse, Arm Cortex-A CPUs, and Automotive Enhanced (AE) IP.
System Interconnects

CMN mesh or CoreLink for optimized bandwidth and scalability.
System IP

Memory controllers, interrupt controllers, SMMUs, and security elements.
Optional Accelerators

Arm Mali GPUs, ISPs, and NPU interfaces based on platform needs.
Firmware and Drivers

Reference software stacks to enable rapid software bring-up.
Implementation Guidance

Floorplanning, power/performance targets, and configurability for silicon partners.

These components are delivered as part of a verified, market-tuned compute foundation that serves as the heart of each Arm platform.

Market Solutions

How Arm Compute Subsystems Are Used

  • Cloud Computing icon
    Cloud and Infrastructure
  • Smartphone
    Mobile
  • Auto icon
    Automotive

Arm Neoverse CSS

Arm Neoverse CSS powers high-performance, energy-efficient SoCs for cloud, AI, and 5G. It enables scalable, customizable infrastructure with consistent architecture across workloads.


It integrates Neoverse CPUs, CMN mesh interconnect, system IP, and supporting software. Built for:

  • Tens to hundreds of cores
  • AI, cloud, telco, and 5G workloads
  • System configurations that cut time-to-silicon by up to 13 months

Adopted by: Amazon, Google, Microsoft, ADTechnology, NeuReality, and XSight Labs

Explore Arm Neoverse CSS
Processor chip on a circuit board symbolizing advanced compute subsystem technology.

Arm Lumex CSS Platform

Arm Lumex CSS unites Armv9 CPUs, Mali GPUs, and advanced process technology to deliver premium performance across smartphones, laptops, and AR/VR—adaptable to any form factor or workload.


It includes Armv9 CPUs, Mali GPUs, physical IP, and cache memory. Key benefits include:

  • Flagship performance and power efficiency on 3nm
  • Customizable for phones, laptops, and AR/VR
  • Already adopted by Samsung (e.g., Galaxy Z Flip 7)
Explore Arm Lumex CSS Platform
Smartphone and laptop displaying AI assistant features across connected devices. (Arm Lumex CSS platform)

Arm Zena CSS

Zena CSS delivers safety-certified compute for SDVs, ADAS, and IVI, integrating safety, security, and AI acceleration to help automakers speed development and standardize platforms.


Built on Armv9-AE technology, Zena CSS features:

  • Cortex-A CPUs, Automotive Enhanced (AE) IP, safety islands, and security enclaves
  • Optional GPU/ISP and support for custom logic
  • ISO 26262 and ISO 21434 compliance
  • Use cases in SDVs, ADAS, and IVI systems
Explore Arm Zena CSS
Arm Zena Compute Subsystem
CSS Successes

Real-World Results with Arm Compute Subsystems

Arm Compute Subsystems (CSS) power some of the world’s most advanced mobile and AI experiences, helping partners accelerate performance, efficiency, and innovation.

Powering Flagship Mobile Experiences

Samsung advances its Exynos chipsets with Arm CSS, delivering performance and power efficiency across the Samsung Galaxy family. Arm now supports the top four Android phone vendors worldwide, showing how the CSS platform enables innovation across the mobile ecosystem.

See Samsung Performance

Accelerating On-Device AI

Google integrates Arm Lumex CSS to power faster on-device inference and real-time personalization in apps like Gmail and YouTube. CSS-based compute brings richer, more efficient AI experiences to billions of Android users worldwide.

Explore Google Performance

Powering Intelligent Financial Experiences

Alipay leverages the Lumex CSS Platform to provide secure, AI-driven personalization across its mobile payment platform. By combining compute efficiency with scalability, CSS helps enhance responsiveness and trust for hundreds of millions of users daily.

Discover Alipay Success

Arm Compute Subsystem (CSS): Frequently Asked Questions

What companies are using CSS today?

Microsoft (Neoverse CSS), Samsung (CSS for Client), and leading automotive OEMs using Zena CSS.

How does CSS reduce time to silicon?

CSS provides verified compute foundations that eliminate the need for bespoke integration and validation, cutting months off development timelines.

Is CSS customizable?

Yes. Partners can tailor memory, accelerators, and system configurations to meet specific product goals.

Why are companies moving to chiplet-based designs?

Chiplet-based designs help overcome the limits of monolithic SoCs as AI workloads scale. By separating compute, memory, and accelerators into modular dies, companies can improve performance-per-watt, reduce design complexity, and iterate faster.

Combined with Arm Compute Subsystems and standardized chiplet interconnects, this approach delivers high-performance AI silicon with lower risk and faster time to market.

What do chiplets actually change compared to traditional SoC design?

Chiplets shift AI silicon design from a single, monolithic chip to a modular approach, where compute, memory, I/O, and accelerators are built and scaled as separate dies. This improves efficiency, eases power and thermal constraints, and enables faster iteration. Combined with Arm Compute Subsystems and Chiplet System Architecture, chiplets deliver high performance with lower cost, risk, and development time than fully custom SoCs.

Key Takeaways

  • CSS powers the platform-first Arm strategy by providing the compute foundation for infrastructure, client, automotive, and emerging markets.
  • Time to market is significantly reduced, with partners reporting up to 12 months faster silicon delivery and millions saved in NRE.
  • Each CSS is pre-integrated and market-tuned, including CPUs, interconnects, system IP, and supporting firmware.
  • Used across real-world deployments, including Microsoft, Samsung, and leading automotive OEMs for SDVs, cloud, and AI-powered devices.
  • CSS is highly configurable and customizable, enabling partners to tailor performance, integration, and features to specific product goals across all markets.