New Arm-based chips bring unified memory and local AI capabilities to Windows PCs, aiming to compete with Apple Silicon performance.
Key points
- RTX Spark combines NVIDIA Grace Arm CPUs with Blackwell GPUs on a single chip.
- The platform supports up to 128GB of unified memory for heavy workloads.
- Microsoft optimized Windows for installing local AI agents via the Store.
Microsoft and NVIDIA introduced the RTX Spark system-on-a-chip for Windows laptops. The hardware aims to bring powerful local AI processing to the PC market.
What happened
Microsoft and NVIDIA held a launch event in San Francisco last week to debut the RTX Spark platform. According to Engadget, the chip powers the Surface Laptop Ultra and other devices arriving this fall. The system integrates up to 20 NVIDIA Grace Arm CPU cores with a Blackwell GPU. It features up to 128GB of unified LPDDR5x memory running at 9,400 Mbps. A lower-tier configuration offers 18 CPU cores and fewer GPU units.
The design includes fifth-generation tensor cores and fourth-generation ray tracing hardware. Microsoft also updated Windows to support local AI agents more easily. Users can install agents like Hermes, Perplexity, and OpenClaw through the Windows Store. Peter Dawoud, a principal program manager at Microsoft, stated that setup requires only two clicks. The Verge reported some installation issues with OpenClaw during testing, requiring a reinstall.
Why it matters
This shift moves AI processing from the cloud to the device itself. Local execution means data stays on the laptop, which can improve privacy and reduce latency for real-time tasks. Unified memory allows the CPU and GPU to share a large pool of fast storage. This architecture eliminates bottlenecks when moving data between components, enabling smoother performance for creative work and complex models.
Teams usually evaluate whether local AI benefits their specific workflow before adopting new hardware. Check whether your applications support on-device inference to maximize efficiency. This approach reduces reliance on internet connectivity for basic AI features. It also lowers long-term operational costs by offloading compute from servers. Users should verify that their security tools can monitor local model activity. Understanding this architecture helps in planning future infrastructure upgrades.
The fix
- Verify that your current applications are compatible with Arm-based Windows systems.
- Check system requirements for local AI models before attempting installation.
- Monitor Windows Store updates for agent compatibility patches.
- Review privacy settings for any new AI services you enable.
Why it matters
When vendors introduce new hardware architectures, software compatibility becomes a primary concern for users. Teams usually audit their existing toolchain to ensure it runs correctly on the new platform. Check whether your critical applications have native support or require emulation layers. This transition period often reveals friction points in deployment workflows. Users should prioritize testing in a controlled environment before widespread adoption. It is essential to understand how memory management changes affect performance expectations. This guidance helps organizations plan for smoother transitions to next-generation devices.
What to do and how to stay safe: NVIDIA
- Check if your essential software supports Arm-based Windows environments.
- Review the system requirements for any local AI tools you plan to use.
- Monitor official vendor channels for driver and compatibility updates.
- Test new AI agents in a limited scope before full deployment.
Step-by-step guide: What is password policies and how they protect systems
General security guidance from the Crypt Weekly newsroom. It is not confirmed advice from the organisations named in this story.
Frequently asked questions
What is RTX Spark?
RTX Spark is a new system-on-a-chip from NVIDIA that combines Arm CPUs and Blackwell GPUs for Windows PCs.
Which laptops will use RTX Spark?
The Surface Laptop Ultra and several other Windows PCs launching this fall will feature the chip.
How do users install AI agents on these devices?
Microsoft allows users to install and update AI agents through the Windows Store with a simplified setup process.
