Microsoft Local AI Expands Beyond NPUs with New Aion Models
Microsoft has unveiled Aion 1.0 Instruct and Aion 1.0 Plan SLMs at Build 2026, expanding Windows AI APIs for CPUs, GPUs, and NPUs. This initiative aims to accelerate local AI development, enabling faster, more private, and cost-effective AI experiences directly on Windows devices. The expansion includes speech-to-text for CPUs and NPUs, and video super resolution for CPUs, signifying a shift towards making Windows an AI-native platform with on-device capabilities.
Key Takeaways
- Microsoft unveiled Aion 1.0 Instruct for chat and content generation, and Aion 1.0 Plan for agentic workflows.
- Windows AI APIs are now supported across CPUs, GPUs, and NPUs, moving beyond NPU-centric AI.
- Speech-to-text recognition APIs are expanded to both CPUs and NPUs, increasing device compatibility.
- On-device language models can now leverage capable discrete GPUs for faster processing.
- Video Super Resolution support is extended to CPUs, allowing broader device-based video enhancement.
Why It Matters
Microsoft's strategy to expand local AI capabilities beyond dedicated NPUs signifies a broader market shift towards on-device processing. By making Windows AI APIs compatible with CPUs and GPUs, the company is democratizing access to AI-powered features for a wider install base, reducing reliance on cloud infrastructure. This move addresses growing concerns around data privacy, latency, and operational costs. For developers, this means the ability to build and deploy more responsive and private AI applications without requiring specialized hardware. The critical metric to watch will be user adoption rates of these new on-device AI features and their impact on cloud AI services revenue for Microsoft and from Microsoft.
Read full article at thewincentral.com